ssci

Al

Overdoing Gender: A Test of the Masculine Overcompensation Thesis1

Robb Willer University of California, Berkeley

Christabel L. Rogalin Purdue University, North Central

Bridget Conlon California State University, Fresno

Michael T. Wojnowicz University of Washington

980 AJS Volume 118 Number 4 (

1We thank Stephen Benard, Nora Broege, Da berg, Claude Fischer, Amal Harrati, Michael

© 2013 by The University of Chicago. All r 0002-9602/2012/11804-0004$10.00

This content downloaded from 139.18 l use subject to University of Chicago Press Terms

The masculine overcompensation thesis asserts that men react to mas- culinity threats with extreme demonstrations of masculinity, a prop- osition tested here across four studies. In study 1, men and women were randomly given feedback suggesting they were either masculine or feminine. Women showed no effects when told they were mascu- line; however, men given feedback suggesting they were feminine ex- pressed more support for war, homophobic attitudes, and interest in purchasing an SUV. Study 2 found that threatened men expressed greater support for, and desire to advance in, dominance hierarchies. Study 3 showed in a large-scale survey on a diverse sample that men who reported that social changes threatened the status of men also re- ported more homophopic and prodominance attitudes, support for war, and belief in male superiority. Finally, study 4 found that higher testosterone men showed stronger reactions to masculinity threats than those lower in testosterone. Together, these results support the masculine overcompensation thesis, show how it can shape political and cultural attitudes, and identify a hormonal factor influencing the effect.

INTRODUCTION

The 1964 film Dr. Strangelove, or How I Learned to Stop Worrying and Love the Bomb depicts the initiation of World War III as a great act of over-

January 2013): 980–1022

na Carney, Shelley Correll, Matthew Fein- Hout, Ted Nordhaus, Andrew Penner, Ol-

ights reserved.

2.075.138 on August 31, 2017 17:12:30 PM and Conditions (http://www.journals.uchicago.edu/t-and-c).

The Masculine Overcompensation Thesis

compensation by a sexually impotent general. This line of reasoning is not unusual, appearing not only in popular but also academic ðe.g., Adler ½1910� 1956Þ discourse. Masculine overcompensation is invoked as an explanation for a variety of behaviors from the everyday—men purportedly purchasing sports cars at the onset of “midlife crises”—to the world-changing—as in analyses of Lyndon Johnson’s escalation of the Vietnam War ðFasteau 1974; Kimmel 1996Þ. The masculine overcompensation thesis asserts that men react to mascu-

line insecurity by enacting extreme demonstrations of their masculinity. Men’s pursuit of masculinity in the face of threats is driven by desires to re- cover masculine status both in their own and others’ eyes. The overcom- pensation dynamic is different from mere compensation. Where compen- sation would lead men to behave roughly as they would normally, their behavior apparently unaffected by the threat, overcompensation suggests a dynamic of relatively extreme reaction, over and above men’s behavior in the absence of threats. In this way, men may inadvertently reveal feelings of threat by behaving in a more extremely masculine way than they otherwise would. If true, the thesis implies that extreme, caricatured demonstrations of masculinity among men may in fact serve as tell-tale signs of underlying insecurity, not self-assured confidence. Those men who exhibit the most masculine traits may actually be seeking cover for lurking insecurities, their outsized masculine displays in fact strategic claims at masculine status, ef- forts to pass as something they fear they are not. But while the logic is familiar—the notion that people would act to cover

up their self-perceived deficits is often cited in settings beyond gender iden- tity ðe.g., Willer, Kuwabara, and Macy 2009Þ—it is not known whether the claim is valid. Do men overcompensate in response to gender identity threats? In this article we review the theoretical bases for the masculine overcompensation thesis, establishing the theoretical claims underlying the dynamic. We then test the idea of masculine overcompensation in a series of laboratory experiments and a large-scale survey. Finally, we conclude by discussing implications of our research.

THEORY

The theoretical roots of the masculine overcompensation thesis lie in psy- choanalytic theory, specifically Adler’s notion that men engage in “mascu- line protest” as a response to feelings of inferiority ð½1910� 1956Þ. Masculine

iver Schultheiss, Michael Shellenberger, the AJS reviewers, and the members of Cornell University’s Laboratory Experimental Group for contributions to this article. Direct cor- respondence to Robb Willer, Department of Sociology, University of California, Berke- ley, California 94720. E-mail: [email protected]

981

This content downloaded from 139.182.075.138 on August 31, 2017 17:12:30 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

American Journal of Sociology

Al

overcompensation also derives from Freud’s notion of “reaction formation” ð½1898� 1962Þ, which describes the tendency of individuals to respond to the suggestion that they possess a socially unacceptable trait by enactingits oppo- site, often in the extreme. While empirical support for psychoanalytic theory in general—and Freud’s defense mechanisms in particular—is decidedly spotty, research is generally supportive of reaction formation ðBaumeister, Dale, and Sommer 1998Þ. In perhaps the best-known empirical demonstration of reaction forma-

tion, one study found that more homophobic men actually showed greater sexual arousal ðas indicated by penile circumference measured via a pe- nile plethysmographÞ than less homophobic men when watching videos of homosexual intercourse, though they reported lower levels of arousal ðAdams, Wright, and Lohr 1996Þ. This finding suggests that homopho- bia may be a case of reaction formation for men with same-sex attrac- tion and strong concerns about the social implications of being seen as gay ðsee also Weinstein et al. 2012Þ. The masculine overcompensation the- sis follows a similar line of reasoning: men who fear they have insuffi- cient masculinity overcompensate by enacting extreme masculine behav- iors and attitudes designed to create the impression that they are quite masculine. Here we review two additional lines of theory that together provide a the-

oretical basis for the masculine overcompensation thesis. First, masculinity theory, which argues that masculinity is both more narrowly defined ðmaking masculinity more easily threatenedÞ and socially valued ðmaking men more motivated to recover itÞ than femininity. Second, theories of iden- tity, which argue that individuals tend to react to feedback that threatens valued identities with overcompensation, enacting attitudes and behaviors associated with the identity to a more extreme extent than they would have in the absence of threats.

Masculinity Theory

Masculinity theorists have traditionally argued that, while definitions of masculinity vary across contexts, within a given culture men are typically measured against a monolithic standard of “hegemonic masculinity” ðCon- nell 1983Þ. Hegemonic masculinity describes the most legitimate and re- spected conception of masculinity in a given culture, prescribing a partic- ular set of behaviors and traits that are viewed as most socially desirable in men ðCarrigan, Connell, and Lee 1985; Connell 1987Þ. Because femininity and other masculinities are less valued and respected than hegemonic masculinity, the stakes associated with maintaining a masculine identity that hews closely to this ideal are relatively high for men. While the spe- cific characteristics of hegemonic masculinity vary across groups and con-

982

This content downloaded from 139.182.075.138 on August 31, 2017 17:12:30 PM l use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

The Masculine Overcompensation Thesis

texts ðConnell and Messerschmidt 2005Þ, several aspects are common ðSchrock and Schwalbe 2009Þ, including competitiveness, assertiveness, physical strength, aggression, risk-taking, courage, heterosexuality, and lack of feminine traits. Researchers have emphasized that dominance and control are central to hegemonic masculinity in the American context ðJohnson 2005Þ. In recent years, some masculinity researchers have turned their attention

to the changing definition of masculinity in American culture and to an ex- pansion of what are viewed as acceptable masculine attitudes and behav- iors ðe.g., Anderson 2009Þ. But despite this work most agree that the cultur- al definition of what are acceptable and respected traits is narrower for masculinity than for femininity ðe.g., Schrock and Schwalbe 2009Þ. For ex- ample, within the literature on children and adolescents, it is clear that while cultural expectations for girls have shown a remarkable transformation in recent years, moving toward acceptance of expressions of self typically asso- ciated with masculinity ðAdams and Bettis 2003Þ, there is a much more halt- ing cultural acceptance of behaviors associated with femininity in boys ðKimmel and Mahler 2003; Pascoe 2005Þ. Masculinity theorists emphasize that sensitivity and responsiveness to

masculinity threats are common in men ðe.g., Kimmel 1994Þ. The social pressure to maintain an esteemed masculine gender identity is strong, though it is different from other forms of normative pressure in that total conformity is likely impossible. Indeed, the standards of true masculinity are so exacting as to be virtually unattainable, leading men to continually strive to satisfy them ðConnell 1987, 1995Þ. As a result, though certain circum- stances may be more emasculating than others, the feeling that one is insuf- ficiently masculine is far from an occasional event. Instead, insecurity, feel- ings of emasculation, and the suspicion of inadequate masculinity are ubiquitous for men. These concerns and feelings of deficiency instigate the enactment of masculinity in everyday life. Because true masculinity is nar- rowly defined, esteemed, and unattainable, a strain always exists, and the result of that strain is overcompensation and the continual striving for ever greater masculinity. Kimmel emphasizes the role of men as a sort of “gender police,” describ-

ing homophobia as the fear among men that other men will detect their in- sufficient masculinity ð1994Þ. This fear, however, is itself a source of shame and must be covered up, along with any possible feminine or inadequately masculine characteristics, ideally with bold demonstrations of strength and masculinity: “What we call masculinity is often a hedge against being revealed as a fraud, an exaggerated set of activities that keep others from seeing through us, and a frenzied effort to keep at bay those fears within ourselves . . . the reigning definition of masculinity is a defensive effort to prevent being emasculated” ðKimmel 1994, p. 103Þ.

983

This content downloaded from 139.182.075.138 on August 31, 2017 17:12:30 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

American Journal of Sociology

Al

Empirical research across a variety of cultures supports the view that masculinity is not easily attained but rather must be continually pursued in the face of threats and challenges ðGilmore 1990Þ. Social psychology re- search on “precarious manhood” supports this view of men as plagued by chronic doubts ðBosson et al. 2009Þ. In one study researchers found that stu- dent participants were able to spontaneously report numerous ways that a man could lose his masculinity ðe.g., losing his jobÞ but relatively fewer ways that women could lose femininity ðVandello et al. 2008Þ. Thus, people find it intuitive that men’s masculine status is tenuous and easily undermined but do not view femininity in the same way. One reason that masculinity is easily threatened is that masculine status

is relative and hierarchical ðKimmel 1994Þ. Thus, as one man establishes his masculine standing, he necessarily diminishes the standing of other men. Indeed, one way to establish one’s own masculinity is by insulting that of another man, creating a self-reinforcing dynamic in which men feed the so- cial pressure that compelled their own masculinity striving in the first place. Thus, masculine insecurity may be in a sense contagious, as it creates in men the seeds of its own perpetuation. Past research on gender and masculinity in the field and laboratory high-

lights men’s sensitivity to threats. For example, Macmillan and Gartner found that employed wives of unemployed husbands face a greater risk of domestic abuse, perhaps because their employment constitutes a threat to the masculinity of their spouse ð1999Þ, a finding also shown for relative spousal income levels ðMcCloskey 1996Þ. Further, men who are more eco- nomically dependent on their wives tend to do less housework ðBrines 1994; but see also Bittman et al. 2003Þ, perhaps acting to distance themselves from behaviors seen as feminine. In one experimental study, social psychol- ogists found that men whose masculinity was threatened via bogus feed- back on a gender identity survey were more likely to subsequently sexually harass a female participant in the study ðMaass et al. 2003Þ. This body of research supports masculinity theorists’ contention that men are highly at- tentive to their masculine status, responding to threats by enacting behav- iors associated with masculinity, including aggression and violence. From this literature we cull two main theoretical claims: ð1Þ a narrower

definition exists for what are acceptable and respected masculine traits, as opposed to feminine traits, and ð2Þ masculinity tends to be more respected than femininity ðe.g., Ridgeway 2011Þ. Together, these dynamics lead men to be both vulnerable to masculinity threats, due to the narrow definition of masculinity, and motivated to reclaim a masculine gender identity, due to the prestige attached to it. This is not to say that social expectations do not also compel women to “do gender” ðWest and Zimmerman 1987; Butler 1990; Bordo 1993Þ, and indeed researchers have argued that a corre-

984

This content downloaded from 139.182.075.138 on August 31, 2017 17:12:30 PM l use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

The Masculine Overcompensation Thesis

sponding ideal of “hegemonic femininity” exists ðe.g., Ussher 1997; Krane et al. 2004Þ and that women sometimes strive to reestablish their femininity in the face of threats ðGriffin 1998; Munsch and Willer 2012Þ. But we argue here that women should be less readily threatened by feedback suggesting they lack femininity and less motivated to reclaim a feminine identity. Thus, in the current investigation we expect stronger reactions to gender identity threats among men than among women. We return in the General Discussion section to the issue of feminine overcompensation and under what conditions we expect that it would be most likely.

Theories of Identity

The above theoretical reasoning clarifies why we expect greater responsive- ness to gender identity threats in men than in women, but it does not explain why we predict that men will overcompensate in the face of such threats. We draw upon theories of identity for a characterization of how individuals recover valued identities in the face of disconfirming feedback. Identity the- orists argue that people strive to maintain identities that are deeply held ðBurke and Tully 1977; McCall and Simmons 1978; Stets and Burke 2000bÞ and socially esteemed ðCialdini et al. 1976Þ. Identity theorists often posit a hierarchy of identities of varying strength ðe.g., Stryker 1980Þ, asserting that people are more motivated to act in ways that maintain strongly held and fundamental identities. Identity theories offer insight on how individuals enact and maintain val-

ued identities in the course of interaction. For example, identity control the- ory ðBurke 2004Þ proposes a cybernetic model of the relationships between self-concept, behavior, and situational feedback. In the model, individuals receive information that is relevant to a given, salient identity and assess whether this feedback is consistent with the identity. When discrepancies are detected, they behave in ways designed to bring situational feedback in line with their identity standard. Importantly, many theories of identity would predict that when people receive social feedback that is not consis- tent with a given identity standard, they will enact more extreme versions of behaviors associated with that identity ðBurke 1991; Heise 2007; Burke and Stets 2009Þ. Importantly, under such conditions individuals are ex- pected not merely to compensate but to overcompensate in an effort to re- cover the identity. For example, if a person assuming a role ðe.g., mother, athleteÞ strongly

associated with some characteristic ðe.g., warmth, competitivenessÞ re- ceived feedback that she or he failed to convey this trait in interaction ðe.g., behaving coldly or passivelyÞ, that person would be likely to enact an extreme form of the characteristic in an effort to restore an average presen-

985

This content downloaded from 139.182.075.138 on August 31, 2017 17:12:30 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

American Journal of Sociology

Al

tation of self closer to the desired level. This view of identity maintenance is akin to the workings of a thermostat, which does not maintain a steady tem- perature but instead starts heating cycles that lift a room’s temperature be- yond the desired level upon receiving feedback that the temperature has fallen too far below its target. Thus, individuals responding to feedback that they have fallen short of expectations associated with a valued self- concept are expected to go beyond prototypical behaviors associated with the identity, acting in ways that are extreme versions of the identity in or- der to reclaim it. Nonetheless, such a response can be self-defeating as the feelings of inadequacy motivating the response are detectable in the ex- tremity of the individual’s reaction. In research on identity processes, gender identification as masculine or

feminine drives gender-relevant behaviors, such as men who identify as highly masculine behaving in a more dominant or competitive fashion ðBurke 1989; Stets and Burke 2000aÞ. Gender identities are thought to typically be strongly held ðBurke and Tully 1977Þ. Further, other research on identity maintenance finds that individuals are more motivated to maintain identi- ties that are highly socially valued ðCialdini et al. 1976Þ. As a result, we would expect men to be strongly driven to recover a masculine gender iden- tity when faced with situational feedback suggesting they lack masculinity, more than we would expect women to strive to recover femininity. Further, men’s behavior in the face of threat should not simply be compensatory, but rather overcompensatory. In an effort to project an overall impression of sufficient masculinity for themselves and others, men should enact ex- treme masculine behaviors when they receive feedback suggesting a lack of masculinity.

Testosterone and Masculinity

A mounting body of physiological research has explored relationships be- tween levels of the steroid hormone testosterone and a variety of social be- haviors in humans. Most notably for the present investigation, several stud- ies have linked testosterone levels with masculine behaviors and attitudes related to dominance, aggression, power, risk-taking, and competitiveness ðe.g., Mazur and Booth 1998; Booth et al. 2006; Carney, Cuddy, and Yap 2010Þ. Most relevant to the present investigation, past research finds that testosterone levels are associated with ð1Þ sensitivity to threats and status standing ðvan Honk et al. 1999; Josephs et al. 2006; Terburg, Aarts, and van Honk 2012Þ and ð2Þ the enactment of dominance behaviors ðMazur and Booth 1998; Mehta, Jones, and Josephs 2008Þ. This body of work and the fact that testosterone levels are typically found at significantly higher levels in men than women suggest the possibility that testosterone levels might play a role in masculine overcompensation.

986

This content downloaded from 139.182.075.138 on August 31, 2017 17:12:30 PM l use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

The Masculine Overcompensation Thesis

But how exactly might testosterone levels and overcompensation be linked? Research on testosterone and masculine-typed behaviors like ag- gression and dominance has typically failed to find consistent main effects of testosterone levels on behavior ðRowe et al. 2004; Archer 2006Þ, suggest- ing that the link between testosterone and behavior is more nuanced than often assumed ðBooth et al. 2006Þ. Most research supports the view that tes- tosterone plays a complex role in influencing traits and behaviors associated with masculinity, both in general and in the specific domain of men’s re- sponses to threats and challenges ðKemper 1990; Josephs et al. 2006Þ. One possibility is that men’s testosterone levels might mediate the over-

compensation effect. It could be that testosterone increases when men are faced with a masculinity threat and these higher levels in turn lead to more masculine behaviors. This possibility is consistent with research on the “challenge hypothesis.” Originally developed to explain the role of testoster- one in nonhuman animals ðWingfield et al. 1990Þ, applied to humans the hypothesis argues that threatened men will exhibit increased testosterone levels, which in turn leads to higher levels of aggression and dominance ðArcher 2006; Trumble et al. 2012Þ. Consistent with this, Cohen et al. ð1996Þ found that men who were bumped and insulted in a laboratory ex- periment showed significant testosterone increases and more aggressive be- havior, especially if they were from the American South, where the research- ers argue such actions are culturally understood to be more threatening to a man’s standing. A more likely alternative is that testosterone levels could moderate the

masculine overcompensation effect. It could be the case, for example, that men with higher basal testosterone levels are more responsive to masculin- ity threats. This is suggested by research on “the mismatch effect,” which shows that higher-testosterone men are more concerned with their status and dominance and more sensitive to threats to their standing ðNewman, Sellers, and Josephs 2005; Josephs et al. 2006Þ. For example, research has found that higher-testosterone individuals look longer at pictures of threat- ening faces than those with lower testosterone ðvan Honk et al. 1999Þ. Fur- ther, studies show that when the status of higher-testosterone individuals is challenged, they react with emotional arousal, increased heart rate, and greater focus on status and power concerns ðJosephs et al. 2003; Josephs et al. 2006Þ. Taken together, these findings suggest the possibility that mas- culine overcompensation is more pronounced among higher-testosterone men.

PRESENT RESEARCH

The above-reviewed theory and research suggests the plausibility of our central claim. On the one hand, masculinity theorists argue that masculinity

987

This content downloaded from 139.182.075.138 on August 31, 2017 17:12:30 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

American Journal of Sociology

Al

is both narrowly defined and highly socially valued, making men relatively more likely to perceive threats to their masculinity and more motivated to respond to them. Theories of identity explain how individuals strive to maintain deeply held and socially valued identities like masculinity, posit- ing that such identities are maintained via an overcompensation dynamic in which individuals respond to threats with extreme forms of identity- consistent attitudes and behavior. Finally, research has linked testosterone levels with responsiveness to threats and the enactment of masculine traits like dominance, though it remains an open question whether testosterone levels might mediate or moderate the masculine overcompensation dy- namic. Based on the above, we hypothesize that men will react to masculinity

threats with extreme demonstrations of masculinity but that women will be far less affected by corresponding threats to their femininity. We test this claim across a series of studies. Study 1 is a laboratory experiment in which we tested whether men whose masculinity was threatened would adopt more masculine attitudes. In study 2 we conducted another laboratory ex- periment in an effort to identify what aspect of masculinity men enact through overcompensation. Study 3 tests the masculine overcompensation thesis in a more diverse sample, using a national survey to examine whether men who report greater feelings that social changes threaten the status of men also tend to espouse more masculine attitudes. In each of these studies, we also test for possible effects for women. Finally, in study 4 we returned to a laboratory setting to explore the possible mediating or moderating role that men’s testosterone levels might play in the overcompensation process.

STUDY 1: AN EXPERIMENTAL TEST

We conducted a laboratory experiment as an initial test of the masculine overcompensation thesis. In the study we administered a gender identity survey to men and women and then gave them randomly determined feed- back indicating that they had scored in either the “masculine” or “feminine” range relative to past study participants. We then looked at how this feed- back might affect their responses on subsequent surveys. Specifically we measured participants’ support for the Iraq War, views of homosexuality, and interest in purchasing a sport utility vehicle ðSUVÞ, all views that were considered masculine in the study population.2 Based on the above theoret-

2Past theory and research have linked masculinity with support for war ðConnell 1985; Christensen and Ferree 2008; Messerschmidt 2010Þ, homophobia ðKimmel 1994Þ, and SUV ownership ðBradsher 2002Þ. Conceptions of masculinity, however, vary across

988

This content downloaded from 139.182.075.138 on August 31, 2017 17:12:30 PM l use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

The Masculine Overcompensation Thesis

ical reasoning, we expect men whose masculinity is threatened to express more support for the Iraq War, more negative views of homosexuality, and more interest in buying an SUV. While we predict effects of masculinity threats but weak or no effects for femininity threats among women, we in- clude women in our study to establish that our predicted effects are in fact unique to men. That said, because the dependent measures we employ are selected to capture masculine attitudes, this study offers limited insight on whether and when women would exhibit overcompensation.

Method

Design and Participants.—The study features a 2ðparticipants were men/ womenÞ � 2ðparticipants’ gender identity was threatened/notÞ experimen- tal design. One hundred and eleven undergraduate students ð60 women, 51 menÞ at Cornell University participated in the study for pay plus the op- tion of extra credit in a sociology class. One participant was excluded from analysis because he reported suspicion regarding the gender identity feed- back. Procedure.—Participants were recruited by fliers advertising payment

for participation in a sociology experiment or by announcements in their undergraduate sociology class. After reporting to the lab, participants filled out a demographic questionnaire and a “gender identity survey,” the text of which is given in the appendix. The gender identity survey was the Bem Sex Role Inventory, a battery that asks respondents to indicate how well a series of adjectives ðe.g., competitive, soft-spokenÞ describes their personality ðBem 1974Þ. After this the participants were asked to wait several minutes while the research assistant ostensibly scored the survey. Participants were given results on their gender identity in a sealed envelope with their name printed on it. Envelopes were filled with feedback sheets prior to the session and sealed so that research assistants would be unaware of the experimental condition. Feedback sheets displayed a 0–50 scale of possible scores on the gender

identity survey, as shown in figure 1. The range from 0 to 25 was the mas- culine half of the scale and from 26 to 50 the feminine half. In the middle of each range, brackets indicated the “average male range” and “average fe- male range” for scores in the study. In actuality, the average ranges given

groups ðConnell and Messerschmidt 2005Þ, thus we conducted a pretest ðN560Þ in a sep- arate sample to assess whether these dependent variables were viewed as masculine among members of the population. Participants rated Iraq War support, negative views of homosexuality, and SUV ownership to be more masculine than they were feminine ðP’s < .001Þ and more masculine than opposing war, having positive views of homo- sexuality, or owning one of the other three vehicles tested ðP’s < .05Þ.

989

This content downloaded from 139.182.075.138 on August 31, 2017 17:12:30 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

FIG. 1.—Format for gender identity feedback provided to participants

American Journal of Sociology

Al

on the feedback sheets, as well as the participants’ scores, were false and created purely for the purposes of manipulating gender identity feedback.3

Men and women in the study were randomly assigned to receive either masculine or feminine feedback. Half of men were told they scored an “11” on the gender identity survey, a score corresponding to the middle of the masculine distribution noted on the feedback sheets, while the other half were given a score of “32,” just inside the feminine range of the scale. Half of the women in the study were told that they scored a “39” on the gender iden- tity survey, corresponding to the middle of the feminine distribution, while the other half were told they scored an “18,” a number just within the typical masculine range of the scale.

Dependent measures.—Participants were next asked to fill out two sur- vey packets. One was a Political Views Survey and assessed participants’ attitudes toward homosexuality and the Iraq War, in addition to a variety of filler items. Participants rated on 7-point scales their support for a consti- tutional amendment banning same-sex marriage, the gay rights movement ðreverse-scaledÞ, and whether they thought homosexuality was “always” or “never wrong.” These three measures were averaged to form a composite

3This notion that masculinity and femininity lie at two ends of a single spectrum of gen- der identity is inconsistent with much of the gender literature and with the actual scoring of the Bem Sex Role Inventory. Nonetheless, pilot testing indicated that participants generally viewed this feedback as highly credible and consistent with their own view of gender identity.

990

This content downloaded from 139.182.075.138 on August 31, 2017 17:12:30 PM l use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

The Masculine Overcompensation Thesis

measure of negative attitudes toward homosexuality ðCronbach’s a 5 .89Þ. Participants also rated on 7-point scales their support for “President Bush’s decision to invade Iraq” and approval of “President Bush’s handling of the war in Iraq.” These two items were averaged to form a composite measure of support for the Iraq War ðCronbach’s a 5 .93Þ. The political views sur- vey also included a standard measure of positive and negative affect ðthe PANASÞ, asking participants to indicate on a 5-point scale how much each of 20 emotions items described their feelings at the time ðWatson, Clark, and Tellegen 1988Þ. Participants also filled out a Car Purchasing Survey, which included

one-page descriptions of four different Ford automobiles ðan SUV, a mini- van, a sedan, and a coupeÞ, including pictures, engine specifications, and fuel economy information. Participants were asked to carefully examine each vehicle’s description before filling out a survey on the back page assessing attitudes toward each vehicle. Participants indicated on a 10-point scale how “desirable” each vehicle was, how much they would be willing to pay for each one, and which they would be most likely to buy. The order in which participants completed the political views and car purchasing sur- veys was randomized. Finally, participants were sensitively debriefed re- garding the deceptive gender identity feedback, thanked for their participa- tion, and paid. Debriefing carefully followed procedures developed for studies involving false feedback ðRoss, Lepper, and Hubbard 1975; Aronson et al. 1990Þ.

Results

Table 1 gives means for men’s and women’s attitudes toward the Iraq War and homosexuality. Ratings are mostly low relative to each composite scale’s midpoint of “4,” suggesting that participants reported generally positive views of homosexuality and low support for war. Turning to our predictions, we find that men whose masculinity was threatened reported significantly greater support for the Iraq War and more negative views of homosexuality than did men in the study whose masculinity was not threatened ðP’s < .05Þ. Women, however, showed no significant differences across conditions in their reported attitudes. Turningtoourpredictionsregardingvehiclepreferences, menwhose mas-

culinity was threatened reported viewing the SUV as more desirable and re- ported being willing to spend more money to purchase it than did unthreat- ened men. On average, men whose masculinity was threatened reported a willingness to pay $7,320 more for the SUV than did men whose masculinity was not. There were no significant differences between threatened and unthreatened men in their ratings of the other three vehicles ðP’s > .25Þ.

991

This content downloaded from 139.182.075.138 on August 31, 2017 17:12:30 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

TABLE 1 The Effects of Gender Identity Feedback on Men’s and Women’s Support for the

Iraq War, Views of Homosexuality, and Desire to Purchase an SUV

Gender Identity Threatened ðMeanÞ

Gender Identity Not Threatened

ðMeanÞ t Men ðN 5 51Þ: Support for Iraq War . . . . . . . . . . . . 3.64

ð1.85Þ 2.65 ð1.52Þ

2.06*

Negative views of homosexuality . . . . 4.03 ð1.68Þ

2.77 ð1.60Þ

2.70**

SUV desirability . . . . . . . . . . . . . . . . 6.56 ð2.63Þ

4.84 ð3.16Þ

2.09*

SUV pay ðin thousands of dollarsÞ . . . 28.00 ð13.76Þ

20.68 ð10.63Þ

2.10*

Women ðN 5 60Þ: Support for Iraq War . . . . . . . . . . . . 2.52

ð1.59Þ 2.40 ð1.39Þ

.30

Negative views of homosexuality . . . . 2.54 ð1.81Þ

2.20 ð1.52Þ

.80

SUV desirability . . . . . . . . . . . . . . . . 5.20 ð3.03Þ

5.17 ð2.74Þ

.05

SUV pay ðin thousands of dollarsÞ . . . 22.52 ð14.6Þ

25.38 ð19.52Þ

.63

NOTE.—SDs are in parentheses. * P < .05. ** P < .01.

American Journal of Sociology

Al

Women showed no significant differences across conditions in how desir- able they felt any of the vehicles were, nor in how much they were willing to pay for them. Finally, a higher proportion of threatened men ð40%Þ than unthreatened men ð16%Þ reported being most likely to buy the SUV from among the four vehicles ðx2ð1Þ 5. 3.57, P 5 .059Þ; however, we found no difference between threatened ð17%Þ and unthreatened women ð17%Þ. We also explored participants’ reported emotions. These analyses were

largely exploratory. One could imagine that men whose masculinity was threatened might report greater negative affect. Alternatively, threatened men might express lower levels of emotions in general in effort to appear more masculine. Results supported the former prediction as threatened men reported feeling more guilty, ashamed, upset, and hostile than unthreatened men ðP’s < .05Þ. Women showed no differences across conditions with one exception; women given feedback that they were masculine reported feeling higher levels of nervousness than women who were told they were feminine ðP < .03Þ.4

4One would expect one of the 20 items to be significant by chance using a .05 significance standard.

992

This content downloaded from 139.182.075.138 on August 31, 2017 17:12:30 PM l use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

The Masculine Overcompensation Thesis

Discussion

Results of study 1 support the masculine overcompensation thesis. As pre- dicted, men whose masculinity was threatened expressed more masculine at- titudes than did men whose masculinity was not. Threatened men reported greater support for war, more negative views of homosexuality, and greater interest in purchasing an SUV, all attitudes that a pretest confirmed were viewed as masculine in the study population. Threatened men also reported significantly greater negative affect on multiple emotion survey items. In con- trast, women showed no significant effects of the gender identity feedback. One possible concern regarding the results of this study is that random as-

signment might have failed to equalize important characteristics across con- ditions of the study. As one way to test for this concern, we looked at partici- pants’ self-reported liberalism as reported on the demographic questionnaire administered before the experimental manipulation. Men reported effectively the same levels of liberalism in the threatened and unthreatened conditions ðt 5 .87, P 5 .39Þ. If anything, more liberal men were more often assigned to the threat condition, an initial difference that would make the predicted ef- fects for support for war and expressed homophobia less likely. Indeed, in a multivariate analysis of these effects controlling for self-reported liberalism, the effects of threat for men become slightly more statistically significant. Women’sreported liberalismalsodid notdiffer significantlyacrossconditions ðt 5 2.26, P 5 .80Þ. We also analyzed participant’s scores on the Bem Sex Role Inventory. Masculinity and femininity scores for men and women did not vary across conditions ðP’s > .29Þ; nor did they interact with any of the main effects reported here. Thus,wehaveevaluatedseveralalternativeaccountsforthedatabesidethe

masculine overcompensation thesis, including the possibilities that ð1Þ men and women react to gender identity threat with extreme gender-typed behav- iors, ð2Þ men assigned to the masculinity-threat condition were simply more politically conservative, and ð3Þ men assigned to the masculinity-threat condi- tion were more ðor lessÞ masculine ðor feminineÞ. In each case we found no support for these alternative explanations.

STUDY 2: MASCULINITY THREATS AND DOMINANCE

The purpose of study 2 is to develop a more general understanding of what aspect of masculinity men seek to bolster in response to masculinity threats. One possibility suggested by the results of study 1 is that men respond to mas- culinity threats by seeking to reacquire feelings of dominance and power, core characteristics of masculinity ðKimmel 1994Þ. If so, it is possible that threatened men in study 1 sought to reassert their desire and support for dom- inance in advocating aggressive military action, subordination of a popula-

993

This content downloaded from 139.182.075.138 on August 31, 2017 17:12:30 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

American Journal of Sociology

Al

tion perceived as violating traditional gender roles, and expressing a desire to purchase a large, powerful vehicle. To test this reasoning more directly, we replicated the methodology of

study 1, employing new dependent measures, including a standard measure of desire for dominance and support for dominance hierarchies, “social dominance orientation” ðPratto et al. 1994Þ. Past research suggests that so- cial dominance orientation could be implicated in masculine overcompensa- tion. Dominance attitudes tend to be higher in men than women ðSidanius, Pratto, and Bobo 1996Þ. Men who identify more strongly as men hold stron- ger dominance attitudes, while the opposite pattern is observed among women ðWilson 2005Þ. Men higher in dominance attitudes are more respon- sive to masculinity threats ðMaass et al. 2003Þ. In turn, dominance attitudes predict support for war and nonegalitarian political attitudes ðSidanius and Pratto 1999Þ. A secondary purpose of this study is to address a potential alternative ex-

planation for study 1. One commonality among two of the dependent mea- sures used in our first study—support for the Iraq War and opposition to homosexuality—is that each is related to conservatism in the contemporary United States. Thus, it could be that men whose masculinity was threatened in that study adopted more conservative political attitudes, perhaps be- cause of cultural associations between masculinity and political conserva- tism ðe.g., Ducat 2004Þ. To test this possibility we added multiple surveys tapping aspects of political conservatism as dependent measures.

Method

Design, participants, and procedure.—The study features a 2 ðparticipants were men/womenÞ � 2 ðparticipants’ gender identity was threatened/notÞ experimental design. One hundred undergraduate students ð60 women, 40 menÞ at Cornell University participated in the study for pay, plus extra credit in a psychology class. Study procedures were identical to study 1 ex- cept that participants completed different dependent measures after receiv- ing feedback on their gender identity.

Dependent Measures

Because it provides an established measure of dominance attitudes, we ad- ministered the “group-based dominance” subscale of a standard survey of social dominance orientation ðPratto et al. 1994; Jost and Thompson 2000Þ. Participants indicated their level of agreement on 10-point scales with a se- ries of statements including, “In getting what your group wants, it is some- times necessary to use force against other groups,” “Superior groups should

994

This content downloaded from 139.182.075.138 on August 31, 2017 17:12:30 PM l use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

The Masculine Overcompensation Thesis

dominate inferior groups,” and “Inferior groups should stay in their place.” Results of this 8-item survey were averaged to form a composite ðCronbach’s a 5 .78Þ. We also included several measures of political conservatism. First, we

asked respondents to indicate on 7-point scales how positively or negatively they felt about a series of political positions: reduced aid to the poor, the U.S. military, higher taxes on the rich ðreverse-scaledÞ, government support for business, the death penalty, and affirmative action ðreverse-scaledÞ. These items were averaged to form an issue-based, 6-item composite measure of political conservatism ðCronbach’s a 5 .63Þ. Second, we measured partici- pants’ levels of system justification, the belief that the status quo is funda- mentally fair, a position strongly associated with political conservatism in past research ðJost, Banaji, and Nosek 2004Þ. Participants indicated on 9-point scales their degree of agreement with a series of statements includ- ing, “Everyone has a fair shot at wealth and happiness” and “In general, you find society to be fair.” Responses to this 8-item battery ðe.g., Kay and Jost 2003Þ were averaged to create a composite ðCronbach’s a 5 .78Þ. Fi- nally, we assessed levels of traditionalism, a core aspect of conservative po- litical beliefs ðJost et al. 2003Þ. Participants indicated on 9-point scales their degree of agreement with statements including, “It’s better to stick with what you have than to keep trying new uncertain things” and “I need to know that something would really work before I’d be willing to take a chance at it.” Their responses to this standard 7-item battery ðMcClosky 1958Þ were aver- aged to form a composite ðCronbach’s a 5 .64Þ.

Results and Discussion

Table 2 shows results for men and women given feedback that was either gender identity threatening or unthreatening. First, we find support for our central prediction. Men whose masculinity was threatened did express stronger dominance attitudes than unthreatened men ðP < .04Þ. However, there was no effect of gender identity feedback on the reported dominance attitudes of women ðP > .40Þ. This finding supports our reasoning that men react to masculinity threats with stronger dominance-related attitudes, in this case reflected in their greater desire for dominance and support for dom- inance hierarchies in society. This finding suggests that increased desire and support for dominance among threatened men may have driven the results of study 1. We found no support, however, for the notion that masculinity threats

simply made men more conservative in general. Threatened men did not re- port more conservative views on a battery measuring attitudes on several political issues, nor greater system justification ðP’s > .25Þ. In fact, threat-

995

This content downloaded from 139.182.075.138 on August 31, 2017 17:12:30 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

TABLE 2 The Effects of Gender Identity Feedback on Men’s and Women’s Dominance Attitudes, Political Conservatism, System Justification, and Traditionalism

Gender Identity Threatened ðMeanÞ

Gender Identity Not Threatened ðMeanÞ t

Men ðN 5 40Þ: Dominance attitudes . . . . . . 4.36

ð1.53Þ 3.26 ð1.61Þ

2.22*

Political conservatism . . . . . 3.77 ð1.07Þ

3.41 ð.96Þ

1.12

System justification . . . . . . . 5.06 ð1.43Þ

5.04 ð1.08Þ

.05

Traditionalism . . . . . . . . . . 2.23 ð.95Þ

2.96 ð1.23Þ

2.11*

Women ðN 5 60Þ: Dominance attitudes . . . . . . 2.91

ð1.43Þ 2.57 ð1.78Þ

.80

Political conservatism . . . . . 3.07 ð.73Þ

3.27 ð.61Þ

1.15

System justification . . . . . . . 4.45 ð1.00Þ

4.55 ð.91Þ

.41

Traditionalism . . . . . . . . . . 2.72 ð.93Þ

2.65 ð.95Þ

.31

NOTE.—SDs are in parentheses. * P < .05. ** P < .01.

American Journal of Sociology

Al

ened men reported significantly less traditionalism ðP < .05Þ, a result we did not expect. Revisiting the items used to assess this construct, it may have been that men whose masculinity was threatened were disinclined to show deference to the ideas of those who came before them, preferring to go with newer, riskier ideas of their own. If so, this finding could be consistent with the notion that men sought to reclaim masculinity by expressing masculine- typed views associated with strength, power, and dominance. We found no effects of gender identity feedback on women’s reported levels of political conservatism, system justification, or traditionalism ðP’s > .25Þ.

STUDY 3: CORRELATIONAL ANALYSES IN A DIVERSE SAMPLE

Studies 1 and 2 demonstrated in laboratory settings that men whose mascu- linity was threatened reacted with more extreme masculine attitudes, in par- ticular views associated with dominance. These studies were, however, con- ducted within a relatively small, homogeneous population, creating external validity concerns ðe.g., Lieberson 1987; but see also Lucas 2003Þ. While laboratory experiments have several unique advantages, for example, ran- dom assignment to controlled conditions sharply reduces spuriousness con-

996

This content downloaded from 139.182.075.138 on August 31, 2017 17:12:30 PM l use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

The Masculine Overcompensation Thesis

cerns, constraining the scope of alternative explanations—the use of demo- graphically homogeneous convenience samples limits confidence in the generality of findings. There may have been something unique to the popula- tion studied that partially drove the observed effects; for example, it could be that young men are uniquely sensitive to threats to their masculinity where older men might not be. Indeed, ample masculinity research suggests that the dynamics of masculinity can vary substantially by race, class, and age ðSchrock and Schwalbe 2009Þ. One way to address these concerns would be to look at the correlation

between masculinity threat and extreme masculine attitudes in a large- scale survey that provides greater demographic heterogeneity. Such a study would offer less certainty that the relationship between the variables was causal, given the threat of spurious and/or reverse causation in correla- tional research, but it would offer much greater confidence that the findings of studies 1 and 2 are externally valid. Specifically, it would reduce concerns that the results of studies 1 and 2 were driven by the use of a sample that was somehow uniquely vulnerable to masculinity threat. In this way, laboratory experiments and correlational analysis of large-scale surveys involving di- verse samples are uniquely complementary methods, as experiments offer le- verage on causal inference lacking in correlational research and large-scale surveys on diverse samples offer the external validity lacking in laboratory experiments. Ideally we would have tested our claims in a large, representative data set,

but we were unable to find one with an adequate measure of our central, in- dependent variable: masculinity threat. We were, however, able to partner with a private research firm on the construction of a large-scale survey fea- turing a diverse sample of Americans, adding to the survey an item gauging perceptions of threat to gender status as well as measures of several of the dependent variables from studies 1 and 2. While the survey was not nation- ally representative and is essentially a convenience sample, it does provide substantial sample diversity. If our results are robust in analyses on this data set, it would provide evidence that the results of studies 1 and 2 were not driven by some unique characteristic of the samples used in those studies. In study 3 we test for relationships between men’s reported feelings that

social changes threaten the status of men and a variety of masculine-typed attitudes. We anticipate that men’s feelings of threat will be related to these attitudes but that there will be no association between feelings of gender threat and the same attitudes among women. In addition to addressing external validity concerns, study 3 is intended to substantively extend the findings of studies 1 and 2. While studies 1 and 2 conceptualized masculin- ity threat as a situational variable, study 3 posits that perceptions that the status of men in the larger society is declining may also threaten men’s

997

This content downloaded from 139.182.075.138 on August 31, 2017 17:12:30 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

American Journal of Sociology

Al

masculinity. Theorists have argued that men may be highly responsive to their perception of the macrolevel status of their gender, being threatened by the rise of feminism, the gay rights movement, and other social trends that could challenge men’s traditional status advantage ðe.g., Kimmel 1996; Schrock and Shwalbe 2009Þ. In addition, laboratory research sug- gests that men’s perception that the status difference between men and women has declined are linked to overcompensatory behaviors, like sexual harassment ðe.g., Dall’Ara and Maass 2000; Maass et al. 2003Þ. Study 3 also extends our study of what masculine-typed attitudes and be-

haviors can be shaped by feelings of threat. In study 3 we again look at the relationship between gender identity threat and support for the Iraq War, views of homosexuality, and dominance attitudes. We also now look at an- other attitude—belief in male superiority—that theorists have argued is closely tied to men’s maintenance of a masculine gender identity ðe.g., Kim- mel 1994; Beneke 1997Þ.

Method

Data from the 2007 American Values Survey ðAVSÞ were acquired from American Environics, a private, California-based research firm. The AVS is a hybrid Internet and mail-based, national survey, conducted every 2–4 years since 1992. It tracks data on an array of values, attitudes, and behaviors over time for commercial and political consulting purposes. Re- spondents were contacted in six waves from a database of records ob- tained from Genesys Sampling. The first two waves of recruitment tar- geted a representative sample of Americans stratified by gender, race, age, and state of residence with a purposeful oversample of California res- idents. Ensuing waves targeted demographic groups underrepresented in the prior waves. In all approximately 2.7% of contacted respondents con- sented to participate, a low response rate by standards of academic re- search, but less so for research in the private sector ðe.g., Putnam 2000Þ. We discuss possible concerns regarding sample selection bias below in the study Discussion. Table 3 gives demographic data on the sample. While di- verse, the sample’s demographic make-up diverged from the U.S. popula- tion in various ways ðe.g., older, more white, more college-educatedÞ and thus should not be taken as representative.

Measures.—We used a variety of demographic variables as controls in analysis: gender, race, age, income, and education. Race and ethnicity were controlled for via dummy variables for whether respondents indicated that they were white, African-American, Latino and/or Hispanic, Asian, or “other.” Education was controlled for with binary variables for whether the respondent’s highest level of education achieved was less than high

998

This content downloaded from 139.182.075.138 on August 31, 2017 17:12:30 PM l use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

TABLE 3 Descriptive Statistics on Control Variables for Study 3

Variable Name Mean

Gender ðfemale/male; %Þ . . . . . . . . . . . . . 46.9/53.1 Age ðyearsÞ . . . . . . . . . . . . . . . . . . . . . 50.2

ð16.5Þ Race/ethnicity ð%Þ: White . . . . . . . . . . . . . . . . . . . . . . . . 79.7 Black . . . . . . . . . . . . . . . . . . . . . . . . 6.2 Latino/Hispanic . . . . . . . . . . . . . . . . 10.8 Asian . . . . . . . . . . . . . . . . . . . . . . . . 2.8 Other . . . . . . . . . . . . . . . . . . . . . . . . .5

Income ð%Þ: Less than $30,000 . . . . . . . . . . . . . . . 18.7 $30,000–$50,000 . . . . . . . . . . . . . . . . 19.4 $50,000–$75,000 . . . . . . . . . . . . . . . . 20.7 $75,000–$100,000 . . . . . . . . . . . . . . . 15.3 $100,000 or above . . . . . . . . . . . . . . . 25.9

Education ð%Þ: Less than high school . . . . . . . . . . . . 6.3 High school degree . . . . . . . . . . . . . . 16.3 Some college or associate degree . . . . 32.6 Bachelor’s degree or more . . . . . . . . . 44.8

Conservatism . . . . . . . . . . . . . . . . . . . . 4.15 ð1.56Þ

NOTE.—N 5 2,210. SDs are in parentheses.

The Masculine Overcompensation Thesis

school, a high school degree, some college or an associate’s degree, or a college or postgraduate degree. Similarly, we created five binary variables corresponding to different levels of annual household income: less than $30,000, $30,000–$50,000, $50,000–$75,000, $75,000–$100,000, and greater than $100,000. Finally, because political conservatism could influence the re- lationships we test, we controlled for respondents’ political ideology based on where they placed themselves on a 7-point scale ranging from “very liberal” to “very conservative.” Basic descriptive statistics and question wording for our key variables are

given in table 4.5 Our central independent variable was Threat to Gender Status, which was based on respondents’ beliefs that societal changes disad- vantaged their gender. Men were asked this question about the status of men, and women about the status of women. Our first dependent variable was support for the Iraq War, which was

based on a single item measure of whether or not the Iraq War was “worth

5Wording of survey items overlaps substantially with those used in studies 1 and 2 be- cause we worked with American Environics in constructing several of the questions in the AVS survey.

999

This content downloaded from 139.182.075.138 on August 31, 2017 17:12:30 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

T A B L E 4

I n d e p e n d e n t a n d D e p e n d e n t V a r i a b l e N a m e s , I n d i v i d u a l I t e m s , D e s c r i p t i v e S t a t i s t i c s , a n d A n s w e r R a n g e s f o r S t u d y 3

V a ri a b le

N a m e

Q u es ti o n W o rd in g a n d A n sw

er C h o ic es

M ea n

M a x /M

in

T h re a t to

g en d er

st a tu s

“R ec en t ch a n g es

in o u r so ci et y o ft en

d is a d v an

ta g e m en

½w om

en �”

ð4 -p o in t sc a le

fr o m

“t o ta ll y d is a g re e”

to “t o ta ll y a g re e” Þ

2 .3 7

ð.9 3 9 Þ

4 /1

S u p p o rt

fo r Ir a q W a r

“A ll in

a ll , d o y o u th in k it w a s w o rt h g o in g to

w a r in

Ir a q , o r n o t? ”

ð4 -p o in t sc a le

fr o m

“t o ta ll y d is a g re e”

to “t o ta ll y a g re e” Þ

2 .1 7

ð1 .1 6 Þ

4 /1

N eg a ti v e v ie w s o f h o m o se x u a li ty

“S o ci et y sh o u ld

re g a rd

p eo p le

o f th e sa m e se x w h o li v e to g et h er

a s

b ei n g th e sa m e a s a m a rr ie d co u p le ” ð4 -p o in t sc a le

fr o m

“s tr o n g ly

a g re e”

to “s tr o n g ly

d is a g re e” Þ

“T o w h a t d eg re e d o th e fo ll o w in g co n tr ib u te

to a m o ra l d ec li n e in

th is co u n tr y ? H o m o se x u a li ty ” ð5 -p o in t sc a le

fr o m

“n o t a t a ll ”

to “a

g re a t d ea l” Þ

½C ro n b a ch ’s

a 5

.8 0 �

.6 3 6

ð.2 8 1 Þ

1 /0

B el ie f in

m a le

su p er io ri ty

“T h e fa th er

o f th e fa m il y m u st

b e th e m a st er

in h is o w n h o m e”

ð4 -p o in t sc a le

fr o m

“s tr o n g ly

d is a g re e”

to “s tr o n g ly

a g re e” Þ

“M en

h a v e a ce rt a in

n a tu ra l su p er io ri ty

o v er

w o m en , a n d n o th in g ca n

ch a n g e th is ” ð4 -p o in t sc a le

fr o m

“s tr o n g ly

d is a g re e”

to “s tr o n g ly

a g re e” Þ

½C ro n b a ch ’s

a 5

.6 2 �

.5 0 4

ð.2 1 0 Þ

1 /0

D om

in a n ce

a tt it u d es

“S o m e g ro u p s o f p eo p le

a re

b et te r th a n o th er s”

ð4 -p o in t sc a le

fr o m

“s tr o n g ly

d is a g re e”

to “s tr o n g ly

a g re e” Þ

“S o m e g ro u p s o f p eo p le

a re

ju st

m o re

w o rt h y th a n o th er s”

ð4 -p o in t sc a le

fr o m

“s tr o n g ly

d is a g re e”

to “s tr o n g ly

a g re e” Þ

“I f ce rt a in

g ro u p s o f p eo p le

st a y ed

in th ei r p la ce , w e w o u ld

h a v e fe w er

p ro b le m s”

ð4 -p o in t sc a le

fr o m

“s tr o n g ly

d is a g re e”

to “s tr o n g ly

a g re e” Þ

½C ro n b a ch ’s

a 5

.6 4 �

.4 8 6

ð.1 8 2 Þ

1 /0

N O T E .—

S D s a re

in p a re n th es es .

This content downloaded from 139.182.075.138 on August 31, 2017 17:12:30 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

The Masculine Overcompensation Thesis

it.” Our measure of negative views of homosexuality was based on a com- posite of items assessing views of gay marriage and whether the respondent felt that homosexuality contributes to “a moral decline in this country.” Our measure of belief in male superiority was also based on a composite, this time of items assessing the degree of respondents’ beliefs that men are inher- ently superior to women and that fathers should be masters of their home. Finally, we measured respondents’ Dominance Attitudes with a composite that included three items from the same measure of social dominance orien- tation used in study 2.

Results

We conducted a series of multivariate analyses of the effect of threat to gen- der status on each dependent variable. We predicted that threat to gender status would be associated with more extreme masculine attitudes among men, but not women. Table 5 gives results of analyses for men. Model 1 was an ordered logit model while models 2, 3, and 4 were ordinary least squares ðOLSÞ regression models. Among the control variables, only political con- servatism had consistent effects on the dependent variables, being signifi- cantly related to the dependent variable in all models presented here. In ad- dition, college-educated respondents tended to show less endorsement of the attitudes we studied. Looking at our predicted effects, across all four models the effect of gender

status threat was significant and in the expected direction for men ðP’s < 05Þ. These effects indicate that the more men felt that the status of their gender was threatened by social changes the more they tended to support the Iraq War, hold negative views of homosexuality, believe in male superiority, and hold strong dominance attitudes. Table 6 gives results of these same analy- ses for women. Here, as predicted, we see that women’s reported feelings of threat were uncorrelated with these dependent variables. These findings show that there was no relationship between women’s feelings that the sta- tus of their gender was under threat and the attitudes we studied ðP’s > .55Þ.6 We also conducted a variety of analyses investigating whether the rela-

tionship between gender status threat and men’s attitudes might vary signif- icantly by race, age, income, or education. Analyses showed no consistent in- teraction effects of threat with these demographic variables. Nonetheless, it is likely that these and other factors can influence the magnitude and char- acter of masculine overcompensation, a potentially fruitful area for future research.

6All multivariate analyses were conducted on the maximum possible respondents, i.e., those respondents who gave responses to all items included in the model. Alternate anal- yses restricting analysis to only respondents who completed all items across all models offered nearly identical results.

1001

This content downloaded from 139.182.075.138 on August 31, 2017 17:12:30 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

TABLE 5 Ordered Logit ðModel 1Þ and Ordinary Least Squares Regression ðModels 2, 3, and 4Þ Coefficients for Models Predicting Men’s Support for Iraq War,

Negative Views of Homosexuality, Belief in Male Superiority, and Dominance

Attitudes in Study 3

Independent Variable

Support Iraq War

ð1Þ

Negative Views of Homosexuality

ð2Þ

Male Superiority

ð3Þ

Dominance Attitudes

ð4Þ Age . . . . . . . . . . . . . . . . 2.007

ð.004Þ .001*** ð.000Þ

2.001** ð.000Þ

.001** ð.000Þ

Race/ethncity: Black . . . . . . . . . . . . . 2.706*

ð.314Þ .088** ð.031Þ

.111*** ð.027Þ

.025 ð.024Þ

Latino/Hispanic. . . . . . . .055 ð.215Þ

.043* ð.022Þ

.062** ð.019Þ

.029 ð.018Þ

Asian . . . . . . . . . . . . . 2.402 ð.365Þ

.067 ð.037Þ

.069* ð.032Þ

.099** ð.029Þ

Other . . . . . . . . . . . . . 2.481 ð.934Þ

2.086 ð.096Þ

2.122 ð.082Þ

.007 ð.077Þ

Income: $30,000–$50,000 . . . . . .090

ð.230Þ .017 ð.024Þ

2.026 ð.020Þ

2.017 ð.019Þ

$50,000–$75,000 . . . . . .240 ð.223Þ

.014 ð.023Þ

2.052** ð.020Þ

2.049** ð.018Þ

$75,000–$100,000 . . . . .237 ð.232Þ

2.020 ð.024Þ

2.029 ð.020Þ

2.032 ð.019Þ

$100,000 or above. . . . .392 ð.216Þ

2.046* ð.022Þ

2.052** ð.019Þ

2.039* ð.018Þ

Education: High school degree . . . .223

ð.286Þ .033 ð.031Þ

2.025 ð.026Þ

2.068** ð.025Þ

Some college . . . . . . . . .106 ð.265Þ

2.032 ð.029Þ

2.038 ð.025Þ

2.085*** ð.023Þ

Bachelors degree or more . . . . . . . . . . . . 2190

ð.265Þ 2.051 ð.029Þ

2.067** ð.025Þ

2.111*** ð.023Þ

Conservatism . . . . . . . . . .806*** ð.047Þ

.091*** ð.004Þ

.033*** ð.004Þ

.009** ð.003Þ

Threat to gender status . . . . . . . . . . . .226***

ð.064Þ .038*** ð.007Þ

.041*** ð.006Þ

.038*** ð.005Þ

Intercept . . . . . . . . . . . . .142** ð.051Þ

.420*** ð.036Þ

.427*** ð.033Þ

F-statistic. . . . . . . . . . . . 40.70*** 19.01*** 9.58*** R2 . . . . . . . . . . . . . . . . . .387 .188 .106 N. . . . . . . . . . . . . . . . . . 1,120 1,157 1,162 1,151

NOTE.—SEs are in parentheses. Omitted categories for race/ethnicity, income, and educa- tion are “white,” “less than $30,000,” and “less than high school,” respectively. * P < .05. ** P < .01. *** P < .001.

This content downloaded from 139.182.075.138 on August 31, 2017 17:12:30 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

TABLE 6 Ordered Logit ðModel 1Þ and Ordinary Least Squares Regression ðModels 2, 3, AND 4Þ COEFFICIENTS FOR MODELS PREDICTING WOMEN’S SQUPPORT FOR IRAQ WAR, NEGATIVE

VIEWS OF HOMOSEXUALITY, BELIEF IN MALE SUPERIORITY, AND DOMINANCE ATTITUDES IN STUDY 3

Independent Variable

Support Iraq War

ð1Þ

Negative Views of Homosexuality

ð2Þ

Male Superiority

ð3Þ

Dominance Attitudes

ð4Þ Age . . . . . . . . . . . . . . . . 2.011**

ð.004Þ .001* ð.000Þ

.001 ð.000Þ

.001* ð.000Þ

Race/ethncity: Black . . . . . . . . . . . . . 2.859**

ð.260Þ .082** ð.027Þ

.121*** ð.022Þ

2.022 ð.019Þ

Latino/Hispanic . . . . . 2.179 ð.207Þ

.007 ð.023Þ

.057** ð.020Þ

.025 ð.017Þ

Asian . . . . . . . . . . . . . 2.435 ð.381Þ

.017 ð.047Þ

.127** ð.039Þ

.059 ð.034Þ

Other . . . . . . . . . . . . . .579 ð.708Þ

.080 ð.102Þ

2.064 ð.079Þ

.084 ð.068Þ

Income: $30,000–$50,000 . . . . . 2.073

ð.189Þ 2.017 ð.021Þ

2.012 ð.018Þ

2.044** ð.015Þ

$50,000–$75,000 . . . . . .218 ð.188Þ

.006 ð.022Þ

2.008 ð.018Þ

2.040* ð.016Þ

$75,000–$100,000 . . . . 2.035 ð.226Þ

2.022 ð.026Þ

.002 ð.022Þ

2.045* ð.019Þ

$100,000 or above. . . . .123 ð.199Þ

2.016 ð.024Þ

2.010 ð.020Þ

2.044* ð.017Þ

Education: High school degree . . . .496

ð.297Þ .081* ð.033Þ

.016 ð.028Þ

2.016 ð.024Þ

Some college . . . . . . . . .234 ð.278Þ

.050 ð.031Þ

.001 ð.026Þ

2.062** ð.022Þ

Bachelors degree or more . . . . . . . . . . . . 2.119

ð.286Þ .007 ð.032Þ

2.072** ð.027Þ

2.079* ð.023Þ

Conservatism . . . . . . . . . .630*** ð.048Þ

.103*** ð.005Þ

.045*** ð.004Þ

.011** ð.004Þ

Threat to gender status . . . . . . . . . . . 2.062

ð.071Þ .012 ð.008Þ

.011 ð.007Þ

.007 ð.006Þ

Intercept . . . . . . . . . . . . .058 ð.047Þ

.244*** ð.039Þ

.436*** ð.034Þ

F-statistic. . . . . . . . . . . . 38.11*** 16.25*** 5.84*** R2 . . . . . . . . . . . . . . . . . .346 .184 .075 N. . . . . . . . . . . . . . . . . . 982 1,026 1,027 1,025

NOTE.—SEs are in parentheses. Omitted categories for race/ethnicity, income, and educa- tion are “white,” “less than $30,000,” and “less than high school,” respectively. * P < .05. ** P < .01. *** P < .001.

This content downloaded from 139.182.075.138 on August 31, 2017 17:12:30 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

American Journal of Sociology

Al

Discussion

Study 3 offers further support for the masculine overcompensation thesis, this time in a more diverse sample than those used in studies 1 and 2. Men who reported that social changes threatened the status of men reported greater support for war, more negative views of homosexuality, stronger be- lief in male superiority, and stronger dominance attitudes. Women, on the other hand, showed no such relationships. Perhaps the most significant concern with the study involves inference

of causation from the correlations we found. These correlations could, for example, be driven by a reverse causal process: a tendency for individuals adopting more traditional masculine attitudes to more readily perceive that the status of masculinity is under threat, perhaps because they endorse a narrower conception of masculinity. Alternatively, some third variable that we failed to control for in our models could drive the observed associations. For example, though we controlled for political conservatism, it could be that individuals with a more traditional cultural worldview are be more inclined to see cultural change as threatening and to adopt traditionally masculine attitudes. We believe these are credible concerns and that the re- lationships we found could be driven in part by these processes. The results of the controlled experiments of studies 1 and 2, however, offer strong ev- idence for a causal effect of perceived masculinity threat on men’s mascu- line attitudes. This experimental evidence helps to address the limited ca- pacity for correlational research to support causal conclusions, providing credibility to our causal claims. Another concern with this study is that the nonrepresentativeness of the

sample and the very low response rate might threaten the validity of the data. These aspects of the data set pose serious concerns for matching the magni- tude of the coefficients and intercepts from our regression models with the “true” effects that exist in the population. However, our purpose here was to test whether the relationship between threat and masculine attitudes would be significant and positive for men ðbut not womenÞ in a more diverse sam- ple than those used in studies 1 and 2. The biggest threat posed by the non- representativeness and the low response rate of the survey we used is the possibility that selection into our sample might somehow be correlated with our predicted effects. Alhough this is possible, we see little reason why a group that is overrepresented in our sample would be uniquely sensitive to masculinity ðbut not femininityÞ threats, thus driving the associations ðand nonassociationsÞ we observe. Further, many of the effects for control vari- ables given in tables 5 and 6 are consistent with past research, that is, the link between conservatism and support for war ðZaller 1991Þ—suggesting that the data set’s validity is not fundamentally compromised by the response rate.

1004

This content downloaded from 139.182.075.138 on August 31, 2017 17:12:30 PM l use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

The Masculine Overcompensation Thesis

One possible alternative explanation for our findings in study 3 is that the rising status of women in educational and professional domains may have led women to report low levels of perceived status threat, impairing our ability to detect any possible effects on gendered attitudes. However, while men did report significantly higher levels of threat ðM 5 2.41Þ than women ðM 5 2.32; t 5 2.36; P 5 .02Þ, the magnitude of the difference was likely too small to support this alternative explanation. These results extend the findings of our prior studies in several ways.

First, they offer external validity for the findings of studies 1 and 2, demon- strating a relationship between men’s feelings of masculinity threat and more extreme masculine attitudes in a diverse sample of Americans studied outside of a laboratory setting. They further suggest that perception that so- cial changes have diminished the status of men may affect men in ways sim- ilar to the individual-level gender identity threats of studies 1 and 2.

STUDY 4: TESTOSTERONE AND OVERCOMPENSATION

The results of studies 1–3 show a close tie between threats to men’s mascu- linity and men’s enactment of more extreme masculine attitudes, especially dominance-related attitudes. As noted above, past research has linked men’s testosterone levels with sensitivity to status threats and dominance behaviors, suggesting the possibility that testosterone might be implicated in the masculine overcompensation process. Here we explore two rival hypotheses suggested by past work regarding

the role of testosterone in masculine overcompensation. First, it is possible that testosterone levels mediate the relationship between threats and ex- treme masculine attitudes, with threats leading to increased testosterone levels and, thereafter, masculine attitudes. Alternatively, basal testoster- one levels could moderate the strength of the effect of threats on masculine attitudes, with men higher in basal testosterone being more prone to over- compensation. To test these possibilities, we returned to a laboratory context. Our study

was largely identical to study 1 except that it was conducted only among men and at several points participants were asked to provide saliva sam- ples. These samples were subsequently used to measure testosterone levels, allowing us to test whether testosterone mediates or moderates the mascu- line overcompensation effect.

Method

Design and participants.—The study features a 2 ðparticipants’ gender identity was threatened/notÞ condition experimental design. Fifty-four un- dergraduate men at the University of Iowa participated in the study for pay.

1005

This content downloaded from 139.182.075.138 on August 31, 2017 17:12:30 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

American Journal of Sociology

Al

Procedure.—Study procedures were largely identical to those of studies 1 and 2. Participants were escorted to testing rooms where they took part in the study individually. As in studies 1 and 2, they first completed a demo- graphic questionnaire and gender identity survey ðBem 1974Þ. They were then given randomly determined feedback on the results of their gender identity survey before being asked to complete a Political and Religious Views Survey and a poststudy questionnaire. At regular intervals throughout the study research assistants asked each

participant to provide a saliva sample in a sterile cryogenic vial. Saliva collec- tion was consistent with standard procedures used in past research ðe.g., Schultheiss, Campbell, and McClelland 2005Þ. Sugarless chewing gum was made available to stimulate participants’ saliva production if necessary ðDabbs 1991Þ. In all, four samples were collected. The first was collected upon completion of the consent form. The second samplewas collected upon receiv- ing the gender identity feedback, though well before any changes in testoster- one levels could be detected in a saliva sample. The third sample was col- lected following completion of the Political and Religious Views Survey. The research assistant delayed returning to the participant’s testing room to assure that the third saliva sample was collected at least 15 minutes after the gender identity feedback was administered, allowing changes in partici- pants’ testosterone levels to be detectable in saliva ðSchultheiss et al. 1999; Schultheiss and Rohde 2002Þ. The fourth sample was collected after com- pletion of the final survey packet. Finally, participants were sensitively de- briefed and paid.

Dependent measures.—We measured participants’ support for the Iraq War and negative views of homosexuality using the same composites as in study 1 ðCronbach’s a’s 5 .88, .92, respectivelyÞ. As a third dependent measure we also added two questions assessing support for President George W. Bush. Support for President Bush could be perceived as a masculine-typed attitude given popular perceptions of both his personality and political positions ðe.g., Ducat 2004; Christensen and Ferree 2008; Messerschmidt 2010Þ. Specifically, President Bush’s support for preemp- tive war, a more permissive torture policy, capital punishment, his oppo- sition to gun control, and other similar positions may have led Americans to associate him with masculine traits, including dominance and aggression. To create a composite measure of Support for President Bush, we asked par- ticipants to indicate on 7-point scales “How much do you approve of the job President Bush is doing in general?” and “How successful do you think Pres- ident Bush has been at his job?” ðCronbach’s a 5 .95Þ.

Testosterone.—Saliva samples were refrigerated at 220° C before being shipped to Salimetrics ðState College, Penn.Þ for assaying and analysis of tes- tosterone levels. All samples were assayed for salivary testosterone using a highly sensitive enzyme immunoassay. Each saliva sample was measured in

1006

This content downloaded from 139.182.075.138 on August 31, 2017 17:12:30 PM l use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

The Masculine Overcompensation Thesis

duplicate to improve reliability. The test had a lower limit of sensitivity of 1.0 pg/ml, an average intra-assay coefficient of variation of 4.6%, and an av- erage inter-assay coefficient of variation of 9.8%. For more details of the en- zyme immunoassay procedure, see Salimetrics ð2010Þ. We classified testosterone levels in the first two saliva samples as pre-

manipulation and levels in the latter two samples as postmanipulation. Levels from the two assays of each of the first two samples were averaged to form a composite measure of premanipulation basal testosterone levels ðCronbach’s a 5 .95Þ, and the two assays of the latter two samples were averaged to create a postmanipulation testosterone composite ðCronbach’s a 5 .96Þ.

Results

Mediation.—The mediation hypothesis predicts that men whose mascu- linity is threatened will exhibit increased testosterone, and this higher testos- terone will in turn lead to more masculine attitudes. As a first step in evaluat- ing the mediation claim, we tested to see if the masculinity threat had a significant effect on testosterone levels. We found no significant difference be- tween the postmanipulation testosterone levels of threatened participants ðM 5 109.54 pg/mlÞ and those of unthreatened participants ðM 5 119.28 pg/ml; tð54Þ 5 1.35, P > .15Þ.7 We also tested the effect of masculinity threat on post- manipulation testosterone levels while controlling for premanipulation tes- tosterone levels in a multiple regression analysis. Again, we found no signif- icant effect of threat on subsequent testosterone levels ðP > .65Þ. Moderation.—Finding no evidence for mediation, we turn next to the

possibility that basal testosterone levels might moderate the masculine over- compensation effect. To test for moderation, we first centered our measure of participant’s basal testosterone ðAiken and West 1991Þ. To test whether participants’ of different testosterone levels responded differently to mascu- linity threats, we conducted a series of multiple regression analyses entering basal testosterone level, masculinity threat, and the interaction of the two as predictors of support for the Iraq War, negative views of homosexuality, and support for President Bush.8 Table 7 gives results of these analyses.

7Because changes in levels of testosterone might not be detectable for longer than we an- ticipated, or might attenuate, we also tested for effects of the masculinity threat in the third and fourth samples individually and found none ðP’s 5 .20, .21, respectivelyÞ. 8Past research on the moderating role of testosterone has treated it as either a continuous ðMehta et al. 2008Þ or dichotomous ðe.g., Newman et al. 2005; Josephs et al. 2006Þ vari- able. Accordingly, we explored alternate versions of the regression models given here in whichtestosterone was either trichotomized or dichotomized in various ways ðe.g.,median split, highest third of distribution vs. lowest two-thirdsÞ. Bayesian information criterion ðBICÞ statistics for these models indicated that the continuous analyses reported here fit the data as well as, or better than, any of these alternatives.

1007

This content downloaded from 139.182.075.138 on August 31, 2017 17:12:30 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

TABLE 7 Ordinary Least Squares Regression Coefficients for Models Predicting Support

for Iraq War, Negative Views of Homosexuality, and Support for President

Bush in Study 4

Independent Variable Support Iraq War

ð1Þ Homophobia

ð2Þ

Support for President Bush

ð3Þ Basal testosterone. . . . . . . . 2.006

ð.007Þ 2.004 ð.010Þ

.005 ð.007Þ

Masculinity threat . . . . . . . . .6331

ð.357Þ .581 ð.558Þ

1.07** ð.375Þ

Threat � testosterone . . . . . .028** ð.010Þ

.0281

ð.016Þ .027* ð.011Þ

Intercept . . . . . . . . . . . . . . . . 2.56*** ð.247Þ

2.98*** ð.386Þ

2.51*** ð.260Þ

F-statistic. . . . . . . . . . . . . . 3.83* 1.60 4.86** R2 . . . . . . . . . . . . . . . . .187 .087 .226

NOTE.—N 5 54. 1 P < .10. * P < .05. ** P < .01. *** P < .001.

American Journal of Sociology

Al

Looking first at participants’ support for the Iraq War, we found a non- significant main effect of basal testosterone level. We also found a margin- ally significant, positive effect of masculinity threat, indicating that at the mean level of basal testosterone, threatened men reported greater support for the Iraq War. Most relevant to the moderation hypothesis, the interac- tion of masculinity threat and testosterone level is positive and significant. These relationships are depicted in figure 2, which shows that responses to masculinity threats were more pronounced among higher-testosterone men. While higher-testosterone men responded to masculinity threats with greater support for the Iraq War, lower-testosterone men whose masculin- ity was threatened actually tended to report lower levels of support for the war. Turning to participants’ views of homosexuality, here we find nonsignif-

icant effects of both basal testosterone level and masculinity threat. The in- teraction of these terms, however, was again positive and significant. Fig- ure 3 illustrates this interaction effect showing that the effect of masculinity threats on men’s views of homosexuality varied greatly depending on pre- manipulation testosterone levels. The higher were men’s testosterone levels, the more they responded to masculinity threats with negative views of ho- mosexuality. Low-testosterone men, however, showed the opposite pattern, with threat leading to less negative views of homosexuality.

1008

This content downloaded from 139.182.075.138 on August 31, 2017 17:12:30 PM l use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

FIG. 2.—Effects of testosterone and masculinity threat on support for the Iraq War

FIG. 3.—Effects of testosterone and masculinity threat on negative views of homo- sexuality.

This content downloaded from 139.182.075.138 on August 31, 2017 17:12:30 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

FIG. 4.—Effects of testosterone and masculinity threat on support for President Bush

American Journal of Sociology

Al

Finally, analyzing participants’ support for Bush, we found a nonsignif- icant effect of basal testosterone level and a positive, significant effect of masculinity threat. The latter effect indicates that, at the mean level of tes- tosterone level, masculinity threats led participants to report greater support for Bush. The moderation hypothesis was again supported, as the interac- tion of testosterone and threat was significant and positive. As shown in fig- ure 4, at higher levels of basal testosterone, participants were more likely to react to masculinity threats with support for Bush. Higher-testosterone men reported greater support for Bush than their unthreatened counterparts, but lower testosterone men showed no change in support for Bush as a re- sult of threats.

Discussion

While we found no evidence that testosterone mediated the masculine over- compensation effect, in study 4 we found consistent support for a modera- tion effect. The interaction of basal testosterone and masculinity threat was significant for all three dependent measures we investigated: support for the Iraq War, negative views of homosexuality, and support for President Bush. In all cases higher-testosterone men were more likely to react to mas- culinity threats by adopting more masculine attitudes. We found that the effects of testosterone were highly contingent. For ex-

ample, among men whose masculinity was not threatened, we found no main effects of testosterone as higher- and lower-testosterone men indicated

1010

This content downloaded from 139.182.075.138 on August 31, 2017 17:12:30 PM l use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

The Masculine Overcompensation Thesis

roughly equivalent support for the Iraq War, negative views of homosexu- ality, and support for President Bush. By contrast, the significant effects we found for the interaction of masculinity threats and basal testosterone levels indicate that, when masculinity was threatened, basal testosterone levels were positively associated with masculine attitudes. These results fit well with a growing body of research showing that levels of testosterone do not typically have significant main effects but that in combination with social stimuli like masculinity or status threats, they can have quite strong interac- tion effects ðe.g., Kemper 1990; Josephs et al. 2006Þ. In a sense, the effects of testosterone levels must be “activated” by situational factors if they are to affect behavior or attitudes. Taken together with past research, our findings suggest that higher-testosterone men are more attentive and responsive to threats to their status, reacting to them with dominance-related behaviors and attitudes. More generally, these findings are promising for future re- search on the interaction of hormones and social context.9

GENERAL DISCUSSION

Results of four studies offer consistent support for the masculine overcompen- sation thesis. In study 1, men whose masculinity was threatened expressed more masculine attitudes than men whose masculinity was not. Women, however, were apparently unaffected by equivalent feedback. Results of study 2 showed that men whose masculinity was threatened adopted stron- ger dominance attitudes, suggesting that these threats motivate men to reas- sert their support for such hierarchies and their position in them. Study 3 rep- licated these findings in the context of a large-scale survey study, showing that results were robust in a more diverse sample. Results of study 3 fur- ther suggested that perceptions that cultural changes threaten men’s stand- ing may serve as masculinity threats in a way similar to the individual-level threats of studies 1 and 2. Finally, because past research has linked masculinity and dominance

with testosterone levels, study 4 investigated what role testosterone might play in masculine overcompensation. While we found no evidence that tes- tosterone levels mediated the effect, we did find evidence for moderation. Men who were higher in basal testosterone were more responsive to mascu- linity threats, expressing more masculine attitudes following these threats. Low-testosterone men, on the other hand, did not overcompensate. Taken together these findings help establish the validity of masculine

overcompensation. Multiple studies using both laboratory experimental

9These results are the first showing a causal effect of hormone levels on political attitudes. Note that one past study showed an effect of election outcomes on partisans’ testosterone levels ðStanton et al. 2009Þ.

1011

This content downloaded from 139.182.075.138 on August 31, 2017 17:12:30 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

American Journal of Sociology

Al

and correlational field methods supported the claim. These findings offer more convincing support together as experiments are uniquely able to an- swer the weaknesses of correlational field research, establishing causality by controlling for possible confounding variables, thus reducing concerns that observed correlations could be spurious. Conversely, large-scale, cor- relational field research addresses weaknesses of laboratory experiments by testing hypothesized relationships outside the potentially artificial labo- ratory context and investigating whether results are robust in more diverse populations than the convenience samples typical of laboratory studies. In addition, the present research also looked at masculine overcompensation at three different levels of analysis, showing how this interactional process may be implicated in macrolevel patterns of political and cultural attitudes, while itself being shaped by endocrinological factors. More generally, the present research is consistent with theoretical work

on both masculinity and identity. In accordance with identity theories, we found that men did not simply strive to compensate when their masculinity was threatened. Rather, they overcompensated, as if striving to pull their average level of masculinity up to more acceptable levels. Masculinity theorists have typically argued that masculinity is both nar-

rowly defined and more socially valued than femininity, at least in contem- porary American culture. Our results are also consistent with these argu- ments, as men but not women showed consistent responsiveness to threats in our studies. Men not only responded by adjusting their attitudes to be more masculine upon receiving feedback that they lacked masculinity, they also reported greater negative affect ðstudy 1Þ, suggesting greater sensitivity to threat. We found no corresponding tendency for gender identity threats to foster negative emotions in women. Our application of masculinity theory here also suggests that the extent to

which men will overcompensate in the face of masculinity threats is likely to vary across cultures. We expect masculine overcompensation to be most likely and most pronounced in cultural contexts that ð1Þ conceptualize ac- ceptable masculinity in a very narrow way and ð2Þ value masculinity at a rel- atively high level. Where these factors are both low, where masculinity is more broadly defined and valued similarly to femininity, it should be harder to threaten men’s masculinity and they should be less motivated to reclaim it. Future cross-cultural research should explore whether these macrocultural factors do in fact moderate the performance of masculinity in everyday life in the way we theorize. In addition, future research could explore causal links between the definition and valuation of gender identities across social con- texts. For example, it may be that the high valuation of masculinity tends to lead to it being narrowly defined as people struggle to understand why some- one would violate the standards of a highly valued identity.

1012

This content downloaded from 139.182.075.138 on August 31, 2017 17:12:30 PM l use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

The Masculine Overcompensation Thesis

Theorists further argue that striving for masculinity in the face of threats is a near ubiquitous aspect of men’s lives because of the hierarchical nature of masculinity, the esteem accorded to it, and the presence of a narrowly de- fined set of ideal masculine traits. Our results are consistent with the prom- inent role of threat in the maintenance of masculinity. We found evidence linking masculinity threats to diverse outcomes, including domestic and foreign policy views, cultural attitudes, consumption practices, and more general views of hierarchy and inequality between groups. We also found that both situationally induced threats ðstudies 1, 2, and 4Þ and perceptions of the declining status of men in the larger society ðstudy 3Þ were related to the expression of relatively extreme masculine attitudes. We found evidence that the core aspect of masculinity that men enacted in

the face of threats was dominance, a fundamental basis of hierarchy and sta- tus differentiation among men and boys ðe.g., Johnson 2005; Martin 2009Þ. Experimental and field data showed that threatened men expressed both greater desire for standing within, and support for the existence of, domi- nance hierarchies. Further, the effects we found for belief in the superiority of men, derogation of homosexuals, support for war, and desire for a large, powerful vehicle may also reflect such dominance striving. These results fit well with past arguments that masculinity is arrayed hierarchically, with men competing to pursue greater masculine status according to reigning cul- tural definitions of what are valued masculine traits. On such a playing field men must respond to threats and provocations with demonstrations of ag- gression, power, and dominance if they hope to achieve masculine status rel- ative to other men.

Limitations and Future Directions

Still, the present research leaves open important questions that present fruit- ful opportunities for future research. A critical point in our argument is that women will typically react less to femininity threats than men do to mascu- linity threats. But there are other, alternative explanations for the gender asymmetry we observed in our studies. For example, it could be that women are equally responsive to gender identity threats but that our focus on mas- culine attitudes did not allow us to detect women’s efforts to restore a fem- inine gender identity. While plausible, this interpretation cannot account for the fact that threats led men but not women in study 1 to report greater neg- ative affect, consistent with our reasoning that gender identity threats affect men more than women. But this is not to say that feminine overcompensation is impossible. On the

contrary, past theory and research argue that a hegemonic feminine ideal ex- ists for women ðe.g., Ussher 1997; Krane et al. 2004Þ and find that, accord-

1013

This content downloaded from 139.182.075.138 on August 31, 2017 17:12:30 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

American Journal of Sociology

Al

ingly, women sometimes overcompensate in the face of threats to their fem- ininity ðe.g., Griffin 1998Þ. We argue here that, in the American context, fem- inine overcompensation will typically require larger threats than those stud- ied here and will be more likely in cultural contexts where femininity is defined in narrow, traditional ways. Further, our application of masculinity theory suggests that the more

narrowly defined and highly socially valued femininity is in a given cultural context, the more likely feminine overcompensation will be. Future cross- cultural research could explore if feminine overcompensation is more com- mon in cultures that construct acceptable femininity in very specific terms and value it more highly, relative to cultures lower on these two criteria. Another very fundamental issue that this research leaves unresolved is

whether masculine overcompensation is driven more by identity or reputa- tional concerns. When men respond to threats with masculinity striving, are they seeking to restore a fundamental aspect of their self-concept that they deeply value or to restore reputational standing in the eyes of others? The relatively high social esteem associated with masculinity could affect either process, increasing the reputational rewards for men who behave in a mas- culine way and also increasing men’s private valuation of a masculine gen- der identity. The dependent measures used in the present research involved anonymous surveys of masculine-typed attitudes, suggesting it is unlikely that greater masculinity was motivated by a desire to impress others. But de- spite this we think it is most likely that masculine overcompensation is mo- tivated by both identity and status concerns. It could be that responses on our anonymous surveys reflect heuristics developed in public settings where masculine status is at stake. This interpretation was not directly tested here, however, and should be evaluated. For example, as a way to get at the roles of identity versus status concerns, future research could experimentally ma- nipulate whether participants’ gender identity feedback was kept private or made public to several other study participants ðsee Lightdale and Prentice 1994Þ. A possible concern with the present research is that participants in stud-

ies 1, 2, and 4 might have construed the feedback they received about their masculinity as referring more to their sexual orientation than masculinity per se. This alternative interpretation of our results is suggested by past re- search noting that, for young American men, being masculine is deeply in- tertwined with maintaining a heterosexual identity ðPascoe 2005Þ. While possible, we believe that the feedback we gave in our studies was primarily interpreted as referring to participants’ masculinity and femininity. The gender identity test that participants completed ðsee app. AÞ made no refer- ence to sexuality nor did the feedback participants received. In addition, we found convergent results in study 3, which employed a self-report measure

1014

This content downloaded from 139.182.075.138 on August 31, 2017 17:12:30 PM l use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

The Masculine Overcompensation Thesis

of perceived gender identity threat. Nonetheless, it would be a worthwhile line of future research to explore how much masculinity feedback is subjec- tively construed as sexuality feedback. It could also be interesting to inves- tigate the opposite mediating process, that the effect of sexual orientation feedback on men’s subsequent behavior might be mediated by feelings of masculinity threat.

The Role of Testosterone

While we have consistently found no results of gender identity threat for women in the studies reported here, the results of study 4 might lead one to wonder if women high in basal testosterone could also be responsive to gen- der identity feedback, perhaps even being sensitive to the suggestion that they possess feminine traits. A variety of research suggests that links between testosterone and behavior operate in similar ways for men and women ðMa- zur and Booth 1998; Josephs et al. 2003; Newman et al. 2005Þ, and some past studies have shown that basal testosterone moderates responses to status- relevant feedback in both men and women ðe.g., Mehta et al. 2008Þ. The dis- tributions of basal testosterone among men and women are such that basal testosterone levels at the top end of typical distributions for women are simi- lar to the lowest levels seen among men ðSalimetrics 2010Þ, meaning that only women with extremely high basal testosterone ðroughly, the top 5% of the distributionÞ would possess levels akin to the men in study 4 who showed responses to masculinity threats. Thus, a worthwhile avenue for future re- search could be to explore whether women with very high testosterone might be responsive to threatening gender identity feedback in a way analogous to men. This reasoning also suggests an intriguing, alternate theory to the one

presented here. It might be that who engages in overcompensation is a matter of basal testosterone level more than gender, with moderate5 and high-testosterone men ðand perhaps extremely high-testosterone womenÞ exhibiting strong responses to gender identity-threatening feedback and low-testosterone men and most women being largely unaffected. Addition- ally, past research indicates that those high in basal testosterone may be sen- sitive to feedback threatening their status in general ðJosephs et al. 2006; Mehta et al. 2008Þ, not just gender identity feedback. Taken together with the present research, this suggests the possibility that people with relatively higher basal testosterone levels react to low status feedback by enacting at- titudes and behaviors intended to assert or reclaim dominance. This alter- nate theory fits well with contemporary testosterone research, but it diverges in important ways from the theoretical reasoning presented here and merits further empirical examination.

1015

This content downloaded from 139.182.075.138 on August 31, 2017 17:12:30 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

American Journal of Sociology

Al

Is Overcompensation a Biological Process?

One interpretation of the results of study 4 is that masculine overcompensa- tion should be viewed as a biological process, but we would caution against this interpretation. First, the study fits well with other research showing that testosterone levels rarely have main effects on human behavior; more com- mon are findings that testosterone levels interact with social stimuli to affect behavior ðe.g., Newman et al. 2005; Josephs et al. 2006Þ. Further, in our re- search we found main effects of situational threats on masculine behaviors across three studies where testosterone levels were not considered. But, as shown in figures 2–4, we found no differences in masculine attitudes by basal testosterone level among unthreatened men. Testosterone effects were only observed when a situational threat, in a sense, “activated” them. But, more generally, viewing biological and social factors as necessarily

separate and competing to explain zero-sum variance in human behavior reinforces a false and antiquated dichotomy. Social and biological processes are deeply intertwined. Just as we have documented here a physiological factor that influences human behavior, social factors shape biological pro- cesses in important ways. For example, life course events ðArcher 2006Þ and situational stimuli ðSchultheiss et al. 1999; Schultheiss and Rohde 2002Þ shape individuals’ testosterone levels and, as documented here and else- where, these levels in turn interact with social factors in affecting behavior ðJosephs et al. 2006Þ. We hope these results suggest the usefulness of incor- porating hormonal and other physiological measurement in sociological re- search, though the vast majority of sociological research questions can be productively studied without regard for this level of analysis.

CONCLUSION

Because the notion of overcompensation is so common in popular discourse, it may be hard to see that these study results could once have been viewed as counterintuitive. It is not so intuitive, however, for individuals to enact ex- treme behaviors as a cover for their insecurity. Perhaps the most successful way for men to disguise their masculine insecurity would be to behave in the same way as unthreatened men do. In the present research, this would have resulted in no observed differences between threatened and unthreatened men. Instead, men tend to overcompensate, in a sense overdoing gender, and in so doing inadvertently reveal themselves to be sensitive to threats. One implication of this finding is that extreme masculine behaviors may in fact serve as telltale signs of threats and insecurity. Perhaps those men who appear most assuredly masculine, who in their actions communicate strength, power, and dominance at great levels, may actually be acting to conceal underlying concerns that they lack exactly those qualities they strive to project.

1016

This content downloaded from 139.182.075.138 on August 31, 2017 17:12:30 PM l use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

The Masculine Overcompensation Thesis

APPENDIX

Gender Identity Survey

The gender identity survey given to participants in studies 1, 2, and 4 is printed below ðsee Bem 1974Þ. Please answer the questions that follow as well as you can. Do not skip

questions. If you are unsure of an answer, please give the answer that seems best to you.

For each of the following words, please pick a number from the follow- ing scale that best indicates how well you think the word describes yourself:

Never or Almost Never True 1-2-3-4-5-6-7 Always or Almost Always True

Once you have picked the number that best describes yourself, enter it into the blank next to the word and move on.

1. S 3. H 5. C 7. I 9. C

11. A 13. A 15. H 17. L 19. F 21. R 23. S 25. H 27. S 29. U 31. M 33. S 35. E 37. D 39. L 41. W 43. W 45. F 47. G 49. A 51. A 53. D 55. C 57. T 59. G

All use su

elf-reliant _______

This content downloaded from 139.182 bject to University of Chicago Press Terms a

2. Yielding _______

elpful _______

4. Defends own beliefs _______

heerful _______

6. Moody _______

ndependent _______

8. Shy _______

onscientious _______

10. Athletic _______

ffectionate _______

12. Theatrical _______

ssertive _______

14. Flatterable _______

appy _______

16. Strong personality _______

oyal _______

18. Unpredictable _______

orceful _______

20. Feminine _______

eliable _______

22. Analytical _______

ympathetic _______

24. Jealous _______

as leadership abilities _______

26. Truthful _______

ensitive to the needs of others ______

28. Willing to take risks _______

nderstanding _______

30. Secretive _______

akes decisions easily _______

32. Compassionate _______

incere _______

34. Self-sufficient _______

ager to soothe hurt feelings _______

36. Conceited _______

ominant _______

38. Soft spoken _______

ikable _______

40. Masculine _______

arm _______

42. Solemn _______

illing to take a stand _______

44. Tender _______

riendly _______

46. Aggressive _______

ullible _______

48. Inefficient _______

cts as a leader _______

50. Childlike _______

daptable _______

52. Individualistic _______

oes not use harsh language _______

54. Unsystematic _______

ompetitive _______

56. Loves children _______

actful _______

58. Ambitious _______

entle _______

60. Conventional _______

1017

.075.138 on August 31, 2017 17:12:30 PM nd Conditions (http://www.journals.uchicago.edu/t-and-c).

American Journal of Sociology

Al

REFERENCES

Adams, Henry E., Lester W. Wright Jr., and Bethany Lohr. 1996. “Is Homophobia As- sociated with Homosexual Arousal?” Journal of Abnormal Psychology 105:440–45.

Adams, Natalie, and Pamela Bettis. 2003. “Commanding the Room in Short Skirts: Cheering as the Embodiment of Ideal Girlhood.” Gender and Society 17:73–91.

Adler, Alfred. ð1910Þ 1956. “Inferiority Feeling and Masculine Protest.” Pp. 45–52 in The Individual Psychology of Alfred Adler: A Systematic Presentation in Selections from His Writings, edited by Heinz L. Ansbacher and Rowena R. Ansbacher. New York: Basic Books.

Aiken, Leona S., and Stephen G. West. 1991. Multiple Regression: Testing and Interpret- ing Interactions. Thousand Oaks, Calif.: Sage.

Anderson, Eric. 2009. Inclusive Masculinity: The Changing Nature of Masculinities. New York: Routledge.

Archer, John. 2006. “Testosterone and Human Aggression: An Evaluation of the Chal- lenge Hypothesis.” Neuroscience and Biobehavioral Reviews 30:319–45.

Aronson, Elliot, Phoebe Ellsworth, J. Merrill Carlsmith, and Marti Hope Gonzales. 1990. Methods of Research in Social Psychology, 2d ed. New York: McGraw-Hill.

Baumeister, Roy F., Karen Dale, and Kristin L. Sommer. 1998. “Freudian Defense Mechanisms and Empirical Findings in Modern Social Psychology: Reaction Forma- tion, Projection, Displacement, Undoing, Isolation, Sublimation, and Denial.” Journal of Personality 66:1081–1124.

Bem, Sandra L. 1974. “The Measurement of Psychological Androgyny.” Journal of Con- sulting and Clinical Psychology 42 ð2Þ: 155–62.

Beneke, Timothy. 1997. Proving Manhood: Reflections on Men and Sexism. Berkeley and Los Angeles: University of California.

Bittman, Michael, Paula England, Liana Sayer, Nancy Folbre, and George Matheson. 2003. “When Does Gender Trump Money? Bargaining and Time in Household Work.” American Journal of Sociology 109:186–214.

Booth, Allan, Douglas A. Granger, Allan Mazur, and Katie T. Kivlighan. 2006. “Testos- terone and Social Behavior.” Social Forces 85:167–91.

Bordo, Susan. 1993. Unbearable Weight: Feminism, Western Culture, and the Body. Berkeley and Los Angeles: University of California Press.

Bosson, Jennifer K., Joseph A. Vandello, Rochelle M. Burnaford, Jonathan R. Weaver, and S. Arzu Wasti. 2009. “Precarious Manhood and Displays of Physical Aggres- sion.” Personality and Social Psychology Bulletin 35:623–34.

Bradsher, Keith. 2002. High and Mighty: SUVS—The World’s Most Dangerous Vehicles and How They Got That Way. New York: Public Affairs.

Brines, Julie. 1994. “Economic Dependency, Gender, and the Division of Labor at Home.” American Journal of Sociology 100:652–88.

Burke, Peter J. 1989. “Gender Identity, Sex, and School Performance.” Social Psy- chology Quarterly 52:159–69.

———. 1991. “Identity Processes and Social Stress.” American Sociological Review 56: 836–49.

———. 2004. “Identities and Social Structure: The 2003 Cooley-Mead Award Address.” Social Psychology Quarterly 67:5–15.

Burke, Peter J., and Jan E. Stets. 2009. Identity Theory. New York: Oxford University Press. Burke, Peter J., and Judy Tully. 1977. “The Measurement of Role/Identity.” Social Forces

55: 880–97. Butler, Judith. 1990. Gender Trouble: Feminism and the Subversion of Identity. New

York: Routledge. Carney, Dana R., Amy J. C. Cuddy, and Andy J. Yap. 2010. “Power Posing: Brief Non-

verbal Displays Affect Neuroendocrine Levels and Risk Tolerance.” Psychological Science 21:1363–38.

1018

This content downloaded from 139.182.075.138 on August 31, 2017 17:12:30 PM l use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

C

C

C

C

C

—

—

—

C

D

D

D

Fa Fr

G

G

H

Jo

Jo

Jo

Jo

Jo

Jo

K

The Masculine Overcompensation Thesis

Al

arrigan, Tim, Bob Connell, and John Lee. 1985. “Toward a New Sociology of Mascu- linity.” Theory and Society 14:551–604. hristensen, Wendy M., and Myra Marx Ferree. 2008. “Cowboy of the World? Gender Discourse and the Iraq War Debate.” Qualitative Sociology 31:287–306. ialdini, Robert B., Richard J. Borden, Avril Thorne, Marcus R. Walker, Stephen Free- man, and Lloyd Reynolds Sloan. 1976. “ Basking in Reflected Glory: Three ðFootballÞ Field Studies.” Journal of Personality and Social Psychology 34:366–75. ohen, Dov, Richard E. Nisbett, Brian F. Bowdle, and Norbert Schwartz. 1996. “Insult, Aggression, and the Southern Culture of Honor: An ‘Experimental Ethnography.’” Journal of Personality and Social Psychology 70:945–60. onnell, Robert W. 1983. Which Way Is Up? Essays on Sex, Class, and Culture. Sydney: Allen & Unwin. ——. 1985. “Masculinity, Violence, and War.” Pp. 4–10 in War/Masculinity, edited by Paul Patton and Ross Poole. Melbourne: Intervention. ——. 1987. Gender and Power. Stanford, Calif.: Stanford University Press. ——. 1995. Masculinities. Berkeley and Los Angeles: University of California Press. onnell, Robert W. and James W. Messerschmidt. 2005. “Hegemonic Masculinity: Re- thinking the Concept.” Gender and Society 19:829–59. abbs, James M. 1991. “Salivary Testosterone Measurements: Collecting, Storing, and Mailing Saliva Samples.” Physiology and Behavior 49:815–17. all’Ara, Elena, and Anne Maass. 2000. “Studying Sexual Harassment in the Labora- tory: Are Egalitarian Women at Higher Risk?” Sex Roles 41:681–704. ucat, Stephen. 2004. The Wimp Factor: Gender Gaps, Holy Wars, and the Politics of Anx- ious Masculinity. Boston: Beacon. steau, Marc F. 1974. The Male Machine. New York: McGraw-Hill. eud, Sigmund. ð1898Þ 1962. “The Neuro-Psychoses of Defense.” Pp. 43–70 in The Stan- dard Edition of the Complete Psychological Works of Sigmund Freud, vol. 3. London: Hogarth. ilmore, David D. 1990. Manhood in the Making. New Haven, Conn.: Yale University Press. riffin, Pat. 1998. Strong Women, Deep Closets: Lesbians and Homophobia in Sport. Champaign, Ill.: Human Kinetics. eise, David R. 2007. Expressive Order: Confirming Sentiments in Social Actions. New York: Springer. hnson, Allan G. 2005. The Gender Knot: Unraveling our Patriarchal Legacy. Phil- adelphia: Temple University Press. sephs, Robert A., Matthew L. Newman, Ryan P. Brown, and Jeremy M. Beer. 2003. “Status, Testosterone, and Human Intellectual Performance: Stereotype Threat as Sta- tus Concern.” Psychological Science 14:158–63. sephs, Robert A., Jennifer G. Sellers, Matthew L. Newman, and Pranjal H. Mehta. 2006. “The Mismatch Effect: When Testosterone and Status Are at Odds.” Journal of Personality and Social Psychology 90:999–1013. st, John T., Mahzarin R. Banaji, and Brian A. Nosek. 2004. “A Decade of System Jus- tification Theory: Accumulated Evidence of Conscious and Unconscious Bolstering of the Status Quo.” Political Psychology 25:881–919. st, John T., Jack Glaser, Arie W. Kruglanski, and Frank J. Sulloway. 2003. “Polit- ical Conservatism as Motivated Social Cognition.” Psychological Bulletin 129: 339–75. st, John T., and Erik P. Thompson. 2000. “Group-Based Dominance and Opposition to Equality as Independent Predictors of Self-Esteem, Ethnocentrism, and Social Policy Attitudes among African-Americans and European Americans.” Journal of Experimen- tal Social Psychology 36:209–32. ay, Aaron C., and John T. Jost. 2003. “Complementary Justice: Effects of ‘Poor but Happy” and ‘Poor but Honest’ Stereotype Exemplars on System Justification and Im-

1019

This content downloaded from 139.182.075.138 on August 31, 2017 17:12:30 PM l use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

K

K

—

K

K

L

Li

Lu

M

M

M

M

M

M

M

M

M

M

N

Pa

Pr

American Journal of Sociology

10

All us

plicit Activation of the Justice Motive.” Journal of Personality and Social Psychology 85:823–37. emper, Theodore D. 1990. Social Structure and Testosterone. New Brunswick, N.J.: Rutgers University Press. immel, Michael S. 1994. “Masculinity as Homophobia: Fear, Shame, and Silence in the Construction of Gender Identity.” Pp. 119–41 in Theorizing Masculinities, ed- ited by Harry Brod and Michael Kaufman. Newbury Park, Calif.: Sage. ——. 1996. Manhood in America: A Cultural History. New York: Free Press. immel, Michael S., and Matthew Mahler. 2003. “Adolescent Masculinity, Homopho- bia, and Violence: Random School Shootings, 1982–2001.” American Behavioral Sci- entist 46:1439–58. rane, Vikki, Precilla Y. L. Choi, Shannon M. Baird, Christine M. Aimar, and Kerri J. Kauer. 2004. “Living the Paradox: Female Athletes Negotiate Femininity and Muscu- larity.” Sex Roles 50:315–29. ieberson, Stanley. 1987. Making It Count: The Improvement of Social Research and Theory. Berkeley and Los Angeles: University of California Press. ghtdale, Jennifer R., and Deborah A. Prentice. 1994. “Rethinking Sex Differences in Aggression: Aggressive Behavior in the Absence of Social Roles.” Personality and So- cial Psychology Bulletin 20:34–44. cas, Jeffrey W. 2003. “Theory Testing, Generalization, and the Problem of External Validity.” Sociological Theory 21:236–53. aass, Anne, Mara Cadinu, Gaia Guarnieri, and Annalisa Grasselli. 2003. “Sexual Ha- rassment under Social Identity Threat: The Computer Harassment Paradigm.” Jour- nal of Personality and Social Psychology 85 ð5Þ: 853–70. acmillan, Ross, and Rosemary Gartner. 1999. “When She Brings Home the Bacon: Labor-Force Participation and the Risk of Spousal Violence against Women. Journal of Marriage and the Family 61: 947–58. artin, John Levi. 2009. “Formation and Stabilization of Vertical Hierarchies among Adolescents: Towards a Quantitative Ethology of Dominance among Humans.” So- cial Psychology Quarterly 72:241–64. azur, Allan, and Allan Booth. 1998. “Testosterone and Dominance.” Behavioral and Brain Sciences 21:353–63. cCall, George J., and J. L. Simmons. 1978. Identities and Interactions. New York: Free Press. cClosky, Herbert. 1958. “Conservatism and Personality.” American Political Science Review 52:27–45. cCloskey, Lann A. 1996. “Socioeconomic and Coercive Power within the Family.” Gender and Society 10:449–63. ehta, Pranjal H., Amanda C. Jones, and Robert A. Josephs. 2008. “The Social Endo- crinology of Dominance: Basal Testosterone Predicts Cortisol Changes and Behav- ior following Victory and Defeat.” Journal of Personality and Social Psychology. 94:1078–93. esserschmidt, James W. 2010. Hegemonic Masculinities and Camouflaged Politics: Un- masking the Bush Dynasty and Its War against Iraq. Boulder, Colo.: Paradigm. unsch, Christin, and Robb Willer. 2012. “The Role of Gender Identity Threat in Percep- tions of Date Rape and Sexual Coercion.” Violence against Women. 18(10): 1125–46. ewman, Matthew L., Jennifer G. Sellers, and Robert A. Josephs. 2005. “Testosterone, Cognition, and Social Status.” Hormones and Behavior 47:205–11. scoe, C. J. 2005. “‘Dude, You’re a Fag’: Adolescent Masculinity and the Fag Dis- course.” Sexualities 8:329–46. atto, Felicia, James Sidanius, Lisa M. Stallworth, and Bertram F. Malle. 1994. “Social Dominance Orientation: A Personality Variable Predicting Social and Political Atti- tudes.” Journal of Personality and Social Psychology 67:741–63.

20

This content downloaded from 139.182.075.138 on August 31, 2017 17:12:30 PM e subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

Pu

R

R

R

Sa

Sc

Sc

Sc

Si

Si

St

St

—

St

T

T

U

V

va

The Masculine Overcompensation Thesis

Al

tnam, Robert D. 2000. Bowling Alone: The Collapse and Revival of American Com- munity. New York: Simon & Schuster. idgeway, Cecilia L. 2011. Framed by Gender: How Gender Inequality Persists in the Modern World. New York: Oxford University Press. oss, Lee, Mark R. Lepper, and Michael Hubbard. 1975. “Perseverance in Self- Perception and Social Perception: Biased Attributional Processes in the Debriefing Paradigm.” Journal of Personality and Social Psychology 32:880–92. owe, Richard, Barbara Maughan, Carol M. Worthman, E. Jane Costello, and Adrian Angold. 2004. “Testosterone, Antisocial Behavior, and Social Dominance in Boys: Pubertal Development and Biosocial Interaction.” Biological Psychiatry 55:546–52. limetrics, LLC. 2010. Expanded Range Salivary Testosterone Enzyme Immunoassay Kit. State College, Pa.: Salimetrics, LLC. hrock, Douglas, and Michael Schwalbe. 2009. “Men, Masculinity, and Manhood Acts.” Annual Review of Sociology 35:277–95. hultheiss, Oliver C., Kenneth L. Campbell, and David C. McClelland. 1999. “Implicit Power Motivation Moderates Men’s Testosterone Response to Imagined and Real Dominance Success.” Hormones and Behavior 36:234–41. hultheiss, Oliver C., and Wolfgang Rohde. 2002. ”Implicit Power Motivation Predicts Men’s Testosterone Changes and Implicit Learning in a Contest Situation.” Hormones and Behavior 41:195–202. danius, Jim, and Felicia Pratto. 1999. Social Dominance: An Intergroup Theory of So- cial Hierarcy and Oppression. New York: Cambridge University Press. danius, Jim, Felicia Pratto, and Lawrence Bobo. 1996. “Racism, Conservatism, Affir- mative Action, and Intellectual Sophistication: A Matter of Principled Conservatism or Group Dominance?” Journal of Personality and Social Psychology 70:476–90. anton, Steven J., Jacinta C. Beehner, Ekjyhot K. Saini, Cynthia M. Kuhn, and Kevin S. LaBar. 2009. “Dominance, Politics, and Physiology: Voters’ Testosterone Changes on the Night of the 2008 United States Presidential Election.” PLoS One 4 ð10Þ: e7543. ets, Jan E. and Peter J. Burke. 2000a. “Femininity/Masculinity.” Pp. 997–1005 in Encyclopedia of Sociology, rev. ed. Edited by Edgar F. Borgatta and Rhonda J. V. Montgomery. New York: Macmillan. ——. 2000b. “Identity Theory and Social Identity Theory.” Social Psychology Quar- terly 63:224–37. ryker, Sheldon. 1980. Symbolic Interactionism: A Social Structural Version. Menlo Park, Calif.: Cummings. erburg, David, Henk Aarts, and Jack van Honk. 2012. “Testosterone Affects Gaze Aversion from Angry Faces outside of Conscious Awareness.” Psychological Science 23:459–63. rumble, Benjamin C., Daniel Cummings, Christopher von Rueden, Kathleen A. O’Connor, Eric A. Smith, Michael Gurven, and Hillard Kaplan. 2012. “Physical Com- petition Increases Testosterone among Amazonian Forager-Horticulturalists: A Test of the ‘Challenge Hypothesis.’” Proceedings of the Royal Society of London B: Biolog- ical Sciences 279:2907–12. ssher, Jane M. 1997. Fantasies of Femininity: Reframing the Boundaries of Sex. New Brunswick, N.J.: Rutgers University Press. andello, Joseph, Jennifer K. Bosson, Dov Cohen, Rochelle M. Burnaford, and Jonathan R. Weaver. 2008. “Precarious Manhood.” Journal of Personality and Social Psychology 95:1325–39. n Honk, Jack, Adriaan Tuiten, Rien Verbaten, Marcel van den Hout, Hans Koppes- caar, Jos Thijssen, and Edward de Haan. 1999. “Correlations among Salivary Tes- tosterone, Mood, and Selective Attention to Threat in Humans.” Hormones and Be- havior 36:17–24.

1021

This content downloaded from 139.182.075.138 on August 31, 2017 17:12:30 PM l use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

American Journal of Sociology

Al

Watson, David, Lee Anna Clark, and Auke Tellegen. 1988. “Development and Valida- tion of Brief Measures of Positive and Negative Affect: The PANAS Scales.” Journal of Personality and Social Psychology 54:1063–70.

Weinstein, Netta, William S. Ryan, Cody R. DeHaan, Andrew K. Przybylski, Nicole Legate, and Richard M. Ryan. 2012. “Parental Autonomy Support and Discrepancies between Implicit and Explicit Sexual Identities: Dynamics of Self-Acceptance and Defense.” Journal of Personality and Social Psychology 102:815–32.

West, Candace, and Don H. Zimmerman. 1987. “Doing Gender.” Gender and Society 1:125–51.

Willer, Robb, Ko Kuwabara, and Michael W. Macy. 2009. “The False Enforcement of Unpopular Norms.” American Journal of Sociology 115:451–90.

Wilson, Marc S. 2005. “Gender Group Identity as a Moderator of Sex Differences in In- terpersonal and Intergroup Dominance.” Pp. 145–67 in Psychology of Gender Identity, edited by Janice W. Lee. New York: Nova.

Wingfield, John C., Robert E. Hegner, Alfred M. Dufty Jr., and Gregory F. Ball. 1990. “The ‘Challenge Hypothesis’: Theoretical Implications for Patterns of Testos- terone Secretion, Mating Systems, and Breeding Strategies.” American Naturalist 136: 829–46.

Zaller, John. 1991. “Information, Values, and Opinions.” American Political Science Re- view 85:1215–37.

1022

This content downloaded from 139.182.075.138 on August 31, 2017 17:12:30 PM l use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).