CHAPTER 3

Why Women Don’t Speak

This is slavery, not to speak one’s thought.

—EURIPIDES, THE PHOENICIAN WOMEN

NOW THAT WE HAVE LOCATED the sources of women’s relative political inactivity in general, we can home in on what specifically holds women back from exercising their voice in public discourse. We argued in the previous chapter that motivation and opportunity are the places to search. Here we ask, how, specifically, do motivation and opportunity depress—or elevate—women’s authority in discussion? If we can identify the types of motivations and the forms of opportunity that affect women, we can better understand how women can make a full contribution to the public meeting, the lifeblood of democracy.

We begin by reviewing studies of gender in order to identify what differentiates men from women.¹ Our purpose is to identify motivations that could depress women’s participation in discussion relative to men’s. We identify characteristics of gender that tend to weaken people’s sense of entitlement to exercise influence. That is, we explore the notion that women influence less than men because they are less likely to have a key motivation for doing so—a sense of entitlement to authority. In addition, women may tend to participate and influence less than men because they may be socialized to avoid conflict or to maintain social bonds. We then focus on opportunity—how the group structure and dynamic may either cue women’s lower motivation to participate or erase the effects of those motivations. A key aspect of group structure and process is gender composition. We explore what differentiates groups with more men from groups with more women. We derive the hypothesis that when women are a minority, the group dynamic operates to the detriment of women, and when women are a large majority, particularly without any men, the dynamic helps women. But equally important is the effect of institutional arrangements. The hypotheses we derive about gender composition are fully qualified by the effects of the group’s decision rule.² Numbers matter depending on rule; put differently, we argue that procedures and gender composition interact to enlarge or neutralize the problem of women’s lower authority.

DIFFERENCES BETWEEN MEN AND WOMEN

Our general framework for understanding gender and authority begins with gender role theory.³ The foundational idea is parsimonious: society classifies people into the categories of man or woman based on their biological differences and expects each category to engage in particular types of thought, feeling, and behavior.⁴ These expectations are reflected in, and reinforced in, many small but cumulative ways, through daily experience, in the course of countless interactions with other people, in many settings. The expectations are widespread and carry a moralistic charge, and in that sense form a prescription, or a norm of behavior. And so women learn to behave in feminine ways that signal accommodation, and men to engage in behaviors deemed masculine, signaling assertiveness.⁵ Thus gender role theory provides an overarching framework for understanding why women and men walk into formal discussions with different proclivities to engage in what society views as a core masculine behavior: to exercise authority.

Direct evidence for the explanatory power of gender role theory is plentiful in studies of leadership behavior. Women tend to be more reluctant to put forward their opinions or assert their preferences in formal situations. Women are less likely than men to engage in leadership behaviors in group interaction, such as giving opinions and making suggestions.⁶ It is no accident that women tend to shy away from taking overt leadership roles in formal settings. As we detail below, when they do so, they often experience adverse consequences, and these consequences testify to the operation of a powerful norm of a gendered division of authority.

To be clear, we are not arguing that the feminine style is inferior and that women should therefore adopt masculine styles. In fact, there are significant advantages to the individual leader and to the group from the more feminine style of leadership. There are at least two reasons why the masculine style is not necessarily optimal. First, leaders who pursue the kind of leadership for which women are rewarded—a cooperative, inclusive, and caring type of leadership—can be highly effective at meeting the group’s collective goals and simultaneously create a high-quality experience for all individual group members. An environment where subordinates are cultivated and supported, credit is shared, and conflict is handled through honest and open communication can be quite positive for everyone involved. Second, a style that moves beyond assertion and into aggression detracts even from men’s authority under some circumstances (Ridgeway and Diekema 1989). Thus masculine leadership styles are in some ways risky not only for women but also for the men who engage in them because they may veer into excessive dominance and backfire.

However, neither the benefits from the feminine style nor the fact that men’s style may carry a penalty if taken too far, negate the main point. As we will show, women tend to suffer adverse social consequences as women when they attempt to exert authority. These negative effects on women can explain why women, on average, may be less motivated to exercise influence in a discussion. The general conclusion from gender role theory, then, is that society tends to create expectations of lower authority for women.

But in addition to the overarching claim that women learn to avoid displays of authority, gender role theory offers three specific psychological mechanisms that pinpoint the sources of women’s influence gap in group discussion. We proceed by discussing each of three pathways by which gender may matter: confidence, conflict aversion, and orientation to social bonds.

Gender Pathway I: Women May Participate Less Because They Have Less Confidence

It is tempting to conclude that women are equal to men in American society. Women have made enormous strides, as detailed in the previous chapter. But women still differ from men in an important respect. They are much more likely than men to underrate their competence, qualifications, and achievement. And as we will see below, their performance is, in turn, more sensitive to this self-rating—women who underrate their ability also perform less well than women who do not, holding constant actual, objective ability. This gap between the performance of underraters and others is greater among women than men. Thus low self-confidence is a more potent negative force among women than men.⁷ In sum, despite objectively similar ability, the sexes tend to diverge in matters related to confidence, in two ways: (1) in their level of self-confidence, and (2) in the degree to which low confidence and signals of ambiguous performance suppress the performance of achievement-oriented behaviors and authoritative actions.

Especially powerful demonstrations of the gender gap in self-confidence come from studies that measure objective competence. Women tend to underrate their competence when considered against objective indicators of ability (Beyer and Bowden 1997). For example, boys tend to rate their mathematical ability higher than girls do, even given the same objective level of ability (Wigfield, Eccles, and Pintrich 1996). Even in verbal ability, where women are more likely than men to excel, female students rate their ability at the same level as male students do (Pajares 2002). A striking illustration comes from the study of Princeton undergraduates described in chapter 2. At Princeton, women’s GPA is slightly higher than men’s, and women have a higher chance than men of graduating with honors, yet women graduate as they entered—rating their “intellectual self-confidence as compared to the average person your age” lower than men do theirs (Princeton University 2011, 49, 55).

Low self-confidence is not merely a problem for underachieving women, or when it comes to skills that are unfamiliar to women. As the informal study of Yale Law School suggested, women tend to feel less confident than men even at the highest levels of achievement, and in the most verbal of all professions. This illustrative evidence obtains in more rigorous studies in a variety of settings. For example, a study of doctoral students who won prestigious postdoctorate awards found that 70% of the men considered their ability to be above average, but only 52% of the women did so (Babcock and Laschever 2003, 77). The gender gap in confidence reaches to the top of the ability spectrum.

Women’s proclivity to underestimate their ability can be so extreme that it has been dubbed the “impostor syndrome.” Peggy Orenstein, an influential journalist and writer, recalled that when she was writing her senior thesis she “became paralyzed … convinced that my fraudulence was about to be unmasked. … I told [my advisor] of the fears that were choking me. ‘You feel like an impostor?’ she asked. ‘Don’t worry about it. All smart women feel that way’” (Babcock and Laschever 2003, 78). Even highly accomplished women can suffer intense doubts about their ability.

Women are not only generally less likely than men to view themselves as competent and meritorious, they are also specifically less likely than men to view themselves as qualified to engage in leadership roles. This flows from the fact that women’s confidence tends to plummet all the more when it comes to masculine actions and roles stereotypically associated with men (Beyer and Bowden 1997). In one experiment, psychologists Debra Instone, Brenda Major, and Barbara Bunker (1983) assembled students into task groups of four members and assigned one of them to supervise the work of the others. Ahead of time they asked the supervisors to indicate their level of confidence in their ability to supervise. On that self-report, the men scored a third higher than the women. Furthermore, the higher the score, the more that the supervisors attempted to exercise influence over the student workers. In other words, assertive, authoritative behavior was correlated with participants’ feelings of confidence and efficacy. The confidence gender gap may explain the authority gender gap.

In addition to the gender gap in confidence stemming from beliefs about competence, there is a related gender gap in resilience in the face of negative feedback. Women tend to be more sensitive to negative cues about their performance, and even to the absence of positive ones. For example, a study of students in introductory economics classes found that among students who received a grade lower than B in the course, the men were more likely than the women to continue in the economics curriculum (Horvath, Beaudin, and Wright 1992). That is, given the identical low grade, women tend to be less persistent. Men have a stronger set point or anchor of perceived competence, which is less likely to move in response to negative information from the environment.⁸ Similar patterns obtain among Princeton undergraduates. Women’s self-evaluations were more affected by the valence of others’ feedback, perhaps because they are more likely than men to view others’ perceptions as accurate and valuable information about themselves (Roberts 1991).

Compared to men, women tend to need more explicit assurances that they are deserving of praise or reward. Several studies have demonstrated that women tend to have a lower sense of entitlement than men to benefits such as job pay. Women pay themselves less money in laboratory settings than men do and believe they should work longer, harder, better, and more efficiently than men believe they should for a given level of pay (Major, McFarlin, and Gagnon 1984). That is, men tend to have a higher sense of entitlement than women. This gender norm shows up quite early. The same gender gap in self-confidence and entitlement that we see among adults is found in children as young as six years old. In one study, first grade girls awarded themselves between 30 and 78% fewer chocolates than boys gave to themselves, for the same level of performance (Callahan-Levy and Messe 1979). Moreover, girls’ low self-payment correlated with a preference for feminine over masculine occupations; girls who indicated an interest in becoming a nurse or a secretary gave themselves less than girls who preferred to be a firefighter or astronaut (Callahan-Levy and Messe 1979).

Given the gender gap in perceptions of self-worth, it is no surprise that women, in a more vulnerable position to start with, are more susceptible to cues about their value. The psychologists Wayne Bylsma and Brenda Major (1992) conducted an experiment in which women and men were either given direct assurances of their worth, or not. When not given assurances, women felt entitled to lower wages than men did. But the gender gap closed when women were told that they had performed well at the job. Women tend to be less confident than men about their value and thus to be more sensitive than men to the absence of clear positive signals that they are meritorious.

A recent brain imaging study further reinforces the notion that the sexes differ in the proclivity to hold steady despite negative signals about performance. The study exposed subjects to information about their relative performance in IQ test questions. Beforehand, subjects took the same test individually, to establish their baseline performance. Then subjects were asked to answer the same questions, this time as part of a group task. After they answered each question, these subjects were told how their performance ranked relative to others in their test-taking group. Of the twenty-seven subjects, thirteen recovered their original performance level while fourteen did not.⁹ What distinguishes the “breakdowns” from the “resilients”? Gender. The rates of resiliency were 21% for women and 77% for men. As one journalist summarized, “If we think others in a group are smarter, we may become dumber, temporarily losing both our problem-solving ability and what the researchers call our ‘expression of IQ’” (Kishida et al. 2012). Becoming temporarily “dumber” may be a problem that women tend to experience more than men. It is a response to a signal that cues what many women have learned to internalize—that they are not valuable actors in society.

This does not mean that women—and men—who are sensitive to their relative performance lack ability and thus can offer less valuable contributions to the group. In the brain imaging study, the people more likely to experience the performance breakdown were the subjects with a high IQ. In other words, it is the members most valuable to the group and those who can best contribute to its mission whose capacities are lost because of group dynamics that cue women’s sense of inferiority. Group interaction that causes some people to feel inadequate, because their views are not supported or are ignored, may undermine the group’s chief goal. And women are more likely than men to suppress their potential contribution to the group in the presence of negative feedback or in the absence of positive affirmations.

Because it is all about power and authority, politics is an arena where the confidence gap, and its effects, might be especially pronounced. Various studies support the notion that women tend to have less confidence than men in political life and link it to the finding that women shy away from leadership positions.¹⁰ In studies of potential candidates, women report less politically relevant experience than similarly qualified men do (Fox and Lawless 2011; Lawless and Fox 2010). As the authors put it, “Despite comparable credentials, backgrounds, and experiences, accomplished women are substantially less likely than similarly situated men to perceive themselves as qualified to seek office. Importantly, women and men rely on the same factors when evaluating themselves as candidates, but women are less likely than men to believe they meet these criteria” (Fox and Lawless 2011, 59). Confidence is the single biggest factor accounting for the gender gap in the decision to run for office (Lawless and Fox 2010). That is, women are far less likely than men to run for office, even when women and men share high levels of professional experiences that would qualify them for public office, and even when women have good social connections that facilitate running (Fox and Lawless 2011). Confidence affects women’s participation in the leadership positions of politics.

Experimental evidence reinforces the notion that low confidence is a powerful negative motivation for women, overriding the positive effect of high qualifications. That is, even highly qualified women are more reluctant than comparable men to put themselves forward in a competition for the most competent member of the group. Kristin Kanthak and Jonathan Woon simulated the decision to run for office in a lab setting (2011). First they assigned subjects to perform a nongendered task and measured the person’s objective performance. Then they asked participants if they were willing either to volunteer or to run against other group members for the position of the group’s representative. The participants were told that the representative would perform the task on behalf of the group. The researchers found that the person’s objective competence in performing the task had a great deal to do with men’s decision to run to represent the group; but competence on the task had no effect on women’s decision to do so. Furthermore, while women were more reluctant than men to compete, they were no more reluctant than men to volunteer to be the group’s representative. Women were thus especially hesitant to take on a leadership role when it involved direct comparison with others—and the attendant possibility of negative feedback that might accompany such comparative judgments.¹¹ Both experimental and observational evidence thus point to the same conclusion: women are reluctant to compete for a leadership position even when their ability is high; confidence is a key element in the gender gap in leadership and authority.

At issue is not only women’s confidence in general or in their political competence and desire to run for public office but also women’s specific confidence in public speaking. That specific form of confidence may be particularly consequential for participation in public discussion. And here too we see the familiar pattern where women are shyer than men, on average. As one female city council member told an interviewer, “Men have more authority when they speak, not because they know more, but because they’re more comfortable. Women are nervous because people say, ‘what does she know?’” (Beck 2001, 59). While this comment may be rooted in different levels of confidence about political knowledge, it also refers specifically to speaking to an audience. And systematic studies of accomplished women confirm the interview results: Potential female candidates indicate less confidence than similar men do in their public speaking ability (Lawless and Fox 2011).¹²

If women have less confidence than men do while engaging in formal speech, it is not because women have a lower ability to speak. Verba, Schlozman, and Brady’s Citizen Participation Study, based on a nationally representative sample, included a measure of the respondent’s vocabulary. The authors found that women’s vocabulary is better than men’s, to a significant degree (Verba, Schlozman, and Brady 1995, 434). That is, women are not at a loss for words.

And yet studies consistently find that women are less confident than men about public speaking. Women show higher anxiety in the days leading up to an oral presentation (Behnke and Sawyer 2000). Among eighth graders, girls are less comfortable than boys when it comes to asking questions in class (Daly, Kriesler, and Roghaar 1994). Self-reports of general attitudes about communication produce the same pattern (Lustig and Anderson 1990). As an American woman in Conover and colleagues’ cross-national study of public discussion noted, “I’m not going to bring it up because I have not studied it; I’ve not read about it. And I don’t want to be made to look dumb.” Even more succinct is this pithy summary by a British woman: “I’m not that brave” (Conover, Searing, and Crewe 2002, 53). The gender gap in general and in political confidence goes hand in hand with the gender gap in public speaking confidence.

Why do women tend to be less confident and assertive? In part because they are likely to face social sanction for asserting themselves and attempting to exercise authority. Specifically, such actions are seen as violating norms of what society constructs as “feminine,” and women are taught that they are valued primarily for their femininity. In the words of Babcock and Laschever, “An assertive personal style can be a gender-norm violation for a woman” (2003, 86).

A series of experiments relying on trained confederates has established that women who engage in a more assertive style tend to receive lower performance and/or likability ratings than either men who use the identical style or women who use a cooperative, accommodating style.¹³ Men’s ratings are unaffected; that is, men can be assertive without incurring a social cost (Carli 1990). Moreover, the same assertive behavior is more likely to be perceived as aggressive or irrational when performed by female rather than by male leaders (Heilman, Block, and Martell 1995). These effects hold in various settings, including group interactions. When the sociologist Cecilia Ridgeway trained female confederates to interact with other members of decision-making groups, she found that women exercised the most influence when they used a stereotypically defined feminine style: friendly, cooperative, caring, but nonconfrontational (1982, 81).¹⁴ Women who attempt to exercise influence in the overt ways associated with men experience the failure that comes from norm violation—negative evaluation of their abilities and social rejection in the form of dislike.

For example, one experiment assigned either a male or a female leader to a decision-making group. All the leaders were confederates of the investigators. These female and male confederates had been trained to engage in the same leadership style during the group decision-making task. Nevertheless, the participants in the study rated the female leaders as more domineering than their male counterparts and gave them lower ratings on ability and skill. They also exhibited more negative facial expressions toward the female than the male leader, despite professing to have no more negative attitudes toward female leaders (Butler and Geis 1990). That is, when women use explicit markers of influence, they elicit powerful signals of social disapproval. Men do not. It is no surprise that powerful gender norms are typically well internalized, and few women test their boundaries. When a social identity group consistently encounters rejection when using power, it comes to question its competence to do so, and to believe that it lacks authority.¹⁵

The negative effects of women’s violation of the norm are sufficiently powerful that they may be triggered even by the seemingly trivial use of nonverbal signals. A study by Carli, LaFleur, and Loeber (1995) exposed undergraduates to either a woman or a man who made the identical persuasive appeal to an unpopular view. The content was identical. But the gender of the messenger varied, and so did the nonverbal communication style of the speaker; in one version the message was filmed using a “task-oriented” style using frequent eye contact, calm hand motions, and fluid speech, and in the other, the speaker used a “social style” communicating “friendliness and affiliation” by smiling, leaning forward, and using unintrusive body language. Male speakers gained most influence with the task-oriented communication style, but women did best with a social style. Moreover, the reason women succeeded best with the social style is that it best elevated their likability—and likability mattered more to women’s influence than it did to men’s (when the audience was male). Women fare worse when they act confidently than when they do not, because confidence is associated with masculinity.

And that fear of acting confidently, whether conscious or not, works preemptively to silence women and to distort the preferences they express. One experiment found that indeed, women’s fear of being perceived to be “a pushy person” and triggering punishment for “being too demanding” explains why they shy away from acting assertively in financial negotiation (Amanatullah and Morris 2010, 260). In the study, men and women set the same financial goals before negotiation, but women acted on them less assertively—so much so that they lost 20% of their starting salary in just the first round. As Eagly and Carli sum it up, “People penalize women for immodesty more than they penalize men, and women respond by displaying modesty” (2007, 168). Put differently, women perceive the existence of a norm that expects them to act like women; that defines feminine behavior as demure, modest, accommodating, and cooperative; and that defines masculine behavior as assertive and dominant (Eagly and Karau 2002). Women tend to conform to this norm, because when they don’t, they experience adverse social consequences—they are disliked more and respected less (Rudman 1998; Rudman and Fairchild 2004).¹⁶

These studies are often conducted in artificial settings with college students, but the same dynamic can be found in actual settings populated by “real” adults, as recounted by the economist Linda Babcock and the journalist Sara Lasch-ever (2003). In interviews with successful women, the authors discovered a common refrain—the women were aware that they were expected to act deferentially and unassumingly. For example, one senior law partner told them that as a teenager her father admonished her to self-monitor: “Honey, you know you can’t act like a tiger. You have to act like a kitten” (104). Another woman, capturing a common perception, said: “[an assertive woman] can often … come across as a bitch to people” (86).

Most vividly, Babcock and Laschever learned of a firm that provides a “Bully Broads” service to companies seeking to “modify” businesswomen whose assertive style is viewed negatively. The firm attempts to teach these women to be “nicer” and become “ladies first” (Babcock and Laschever 2003, 85). A representative of the firm explains the problem: “a male executive … has more permission to be an ass. But when women speak their minds, they’re seen as harsh” (85). As we have seen above, women face a particular difficulty when they speak their minds. The “Bully Broads” program trains women to speak hesitantly, self-deprecatingly, and apologetically. Most men sent to the program are sent to learn how to reduce stress or to delegate more; by contrast, almost every woman is sent to become “nicer.”

Women are, therefore, socialized by positive and by negative reinforcements; they can observe the negative effects of their masculine actions and the positive effects of their feminine ones on how others in turn treat them. A woman who violates the gender norm may find herself sent to “nice” boot camp under the threat of losing her job—symbolically, in the form of social sanctions, if not literally. And this more anecdotal evidence is seconded by national surveys showing that while women are no longer expected to occupy themselves primarily in the domestic sphere, taking care of their family, they are nevertheless disfavored in positions of power. For example, a national Gallup poll found that Americans who prefer a male boss outnumber those who prefer a female boss by two to one.¹⁷

This reluctance to assert one’s self and one’s preferences lest one violate the norm of feminine behavior extends to various situations where people attempt to reach a formal agreement. Women are less likely to negotiate at all, to negotiate energetically, and to reap the rewards of negotiation (Eagly and Carli 2007, 169; Babcock and Laschever 2003). When researchers asked MBA students to simulate a job interview and then asked the students if they knew their worth during the salary negotiation, they found that 85% of the men indicated that they knew their worth, while 83% of the women indicated that they were unsure of their worth (Baron 2003). Fully 70% of the men believed they were entitled to above-average pay, while only 30% of the women did. As one woman said to the researcher, “it’s hard to talk about yourself … you tend to just be humble and hope that people can see that you’re a great person” (Eagly and Carli 2007, 170).¹⁸ On average, women are not only less sure of the value of their abilities, but also they are more reluctant to communicate that value to others.¹⁹ Women are trained to view self-assertion as self-promotion and are conditioned to avoid both.

In sum, women tend to be less confident, more sensitive to negative signals about their ability, qualifications, and value, and more reluctant to assert themselves in formal situations than men. They are also specifically less confident about their political competence and more shy about expressing their opinions in public than men. Yet they tend to have stronger verbal abilities than men. In general, women’s confidence gap is not explained by lower objective aptitude. The solution to this apparent puzzle is societal norms about gender. Women’s relative reluctance to assert their views in public meetings comes not from ability but from opportunity and motivation. If the conditions of deliberation do not address the gender gap in confidence, they may create a participation and representation deficit for women.

We have dwelt at length on confidence because it is the most direct hypothesis from gender role theory. In addition, many studies support the hypothesis that the gender gap in influence and achievement is rooted in confidence. However, gender role theory is also consistent with two other hypotheses.

Gender Pathway II: Women May Participate Less Because They Dislike Competition or Conflict

Even if politics were a woman’s game, it might not invite women to participate if it was played like a man’s game. Women may be more reluctant than men to engage in a discussion of public matters when it is conflictual. Eagly found that women are more likely than men to use cooperative and collaborative conflict resolution while men tend to take a winner-take-all, competitive approach to conflict (Eagly, Johannesen-Schmidt, and Engen 2003). In a comprehensive review of gender differences in economic behavior, Croson and Gneezy concluded that women are less likely than men to participate in contests, auctions, and bargaining, and they tend to prefer compensation by piecework over compensation that involves competition (2009).

For example, Niederle and Vesterlund ran a laboratory study in which subjects were assembled in groups of four, two men and two women, and had to complete tasks for rewards (2007). In the first round, the participants were given their payments as they completed the tasks, without competition. In a later round, they performed the tasks again, but this time their rewards depended on a competitive tournament. They learned their absolute performance after each task, but not how well they did relative to other participants. Then the subjects were given a choice—to be rewarded by tournament or by piecework. Most men (73%) chose the tournament, and most women did not (35%). The investigators found that one reason for this sizable gender gap is men’s overconfidence in their performance, but the other cause is women’s reluctance to compete even when they perform well. As the authors summarize it, “low-ability men enter the tournament too much, and high-ability women do not enter it enough” (Niederle and Vesterlund 2007, 1069). One factor that did not explain most of the gender gap was actual performance, which was the same for men and women. Again we see that women’s lagging participation is not due to ability.²⁰ While this study confirms the importance of confidence, it also points to competition aversion by most women, and competition seeking by most men.

Two possible explanations for women’s aversion to competition, especially against men, have emerged from existing research. First, unlike men’s performance, women’s performance is rated lower when they engage in competitive rather than accommodating behavior (Croson and Gneezy 2009; Eagly and Carli 2007), suggesting that women’s interests are negatively affected when they emphasize conflict and competition over other alternatives. Second, the gender difference emerges early in life and may reflect long-term patterns of gender socialization. Girls and boys tend to pursue different patterns of play, with boys more likely than girls to engage in competition and conflict (Miller, Danaher, and Forbes 1986; Maccoby 1988).

A further clue that these processes are driven by cultural norms of gendered behavior comes from an intriguing study of gender and competition in two cultures that differ in women’s status and role expectations. Gneezy and colleagues compared gender differences in the decision to enter a competition in two societies, one patrilineal (the Masai in Tanzania) and one matrilineal (the Khasi in India). They found the usual male proclivity to enter competitions in the patrilineal society, but a reversal in the matrilineal society, where women are much more likely than men to enter competitions (Gneezy, Leonard, and List 2009). If women are socialized to be more cooperative and interdependent with others, then women may dislike situations where there is conflict or competition, or even merely a lack of cooperation. When they are in such situations, women may tend to withdraw from the interaction in order to distance themselves from the conflict. The common denominator for these preferences may be the aversion to situations where the social ties of the participants are frayed.

As Hansen puts it, “women tend to be the peace-keepers and consensus-builders” (1997, 79). A number of studies specifically point to the deterring or silencing consequences of conflict for women: “women are especially likely to remain silent if they anticipate opposition.”²¹ Consistent with the reports from Yale Law School and Princeton University, for example, Houston and Kramarae observe that the aggressive, competitive nature of discussion in traditional classrooms discourages female students from contributing (Houston and Kramarae 1991). Recall, too, that in the Kanthak and Woon study we described earlier, women were no less likely than men to volunteer to represent their team, but were less likely than men to choose to run in an election to obtain a representative position. One difference between these two acts revolves around the possibility of competition and conflict. As Atkeson and Rapoport write, “conflict may be particularly problematic for women who are often more interdependent socially and thus less inclined to engage in activities that might ‘rock the boat’” (Atkeson and Rapoport 2003, 499). This is seconded by Susan Beck’s study of local council members. The women interviewed mentioned partisan conflict as their least favorite part of the job (Beck 2001, 62). The practice of accusing political opponents of misdeeds, launching attacks, and denying credit to meritorious opponents out of political strategy, all seemed to these women to be hateful (Beck 2001, 62).

Women may not only tend to be averse to conflict but may perform less well than men under competitive conditions. A study of students at a selective Israeli engineering school that rewarded students for solving difficult puzzles found that as the level of competition in the testing environment increased, male performance improved significantly, but female performance dropped (Gneezy, Niederle, and Rustichini 2003).

In politics specifically there are hints that women tend to prefer consensus modes of interaction. Kathlene (1994) found that female committee chairs in the Colorado legislature tended to use their positions to facilitate open discussion among many of those present, including committee members, witnesses, and bill sponsors. By contrast, male chairs tended to exercise more of their individual agency and ran the hearings to maximize their own preferences. National samples of state legislative leaders yield similar findings. Women are more likely than men to prefer collaboration in lawmaking. Men committee chairs tend to place less emphasis than female chairs do on a process of inclusion and consensus.²² At the city level, unelected city managers exhibit a similar gendered pattern (Fox and Schuhmann 1999).

These findings for political settings echo findings from other settings. In firms and organizations, women tend to prefer and to practice more consensual styles. A large survey of managers found that women are more likely than men to follow managerial practices such as “people development” and “participative decision-making,” while men are more likely than women to use an individualistic decision-making style and seek more control.²³ In general, women tend to use a more democratic and participatory style and to focus on educating and improving those in their charge (Eagly and Carli 2003).

In sum, women may tend to prefer—and to thrive in—conditions of cooperation and to do less well than men in conditions of conflict. The conditions of discussion may well entail more or less cooperation or conflict. Attending to these conditions is, therefore, key because the level of harmony in the discussion may be especially consequential to women’s participation.

Gender Pathway III: Women May Participate Less Because They Are More Sensitive to Social Bonds

Some of the findings we have reviewed so far not only support the notion that women tend to prefer consensus and to dislike conflict, but also that women may tend to be more oriented to social bonds. Disagreement can find more expression—and better resolution—when people feel connected to each other in bonds of cooperation, affection, or acceptance (Mansbridge 1983). As we saw in chapter 1, recent writing in political philosophy argues that in order to raise the quality of discussion, the first item of business is to establish these bonds. There is little productive exchange of reason without cooperation, according to this theory; the social precedes the cognitive. Exchange across lines of political difference, as Diana Mutz (2006) found, depends heavily on the nature of the social relationships in which people are embedded. We speculate that this effect may be particularly important to women.

The broad reason for this hypothesis is that women may tend to be more sensitive to the presence or absence of social ties. The foundational text on this thesis is Carol Gilligan’s In a Different Voice (1982). This book argued that men and women, from a young age, exhibit different orientations to moral thought and action. Most generally, men tend to prioritize and think along lines of rights, and the regulation or freedom of those rights, while women tend to be motivated more by need for help and by their perceived ties to others, and to approach the world with empathy.

While valid criticisms have been launched against that particular book, the general point has received considerable support.²⁴ Other scholars, working from a different theoretical framework, have argued similarly that society tends to socialize women to “affirm their identities as members of a collective by attending to others” (Amanatullah and Morris 2010). Women tend to prioritize relationships in their lives more than men do, and they do so across the life span and regardless of class (Cross and Madson 1997). As we noted above, societies convey the expectation that a person should conform to gender roles. Men are expected and trained to have the attributes that go hand in hand with masculine roles, such as agency, assertion, and independence; women are expected to exhibit caring for and interdependence with others. Some studies support the notion that women tend to be more empathetic than men (Baron-Cohen 2003; Feldman and Steenbergen 2001). Women read nonverbal cues such as facial expressions more accurately (Falk 1997, cited in Hannagan and Larimer 2010), supporting the idea that women are especially attuned to the reactions of others. Women are more affected than men by the social context and by social cues in economic game experiments (Croson and Gneezy 2009). Women also conform to social pressures more than men when individuals meet face to face and are instructed to communicate their preference to other members (Bond and Smith 1996; Cooper 1979; Eagly and Carli 1981). Women score lower than men on Social Dominance Orientation, a stable personality tendency that undergirds a desire to dominate others (Sidanius and Pratto 1999). In Western cultures, women tend to orient toward interdependence and cooperation and to prioritize the principles of equality or need over merit even at the expense of their own self-interest, while men in turn tend to emphasize autonomy and to behave hierarchically.²⁵ Again, we emphasize that these often tend to be small differences between individuals, with variance within each gender, but may be amplified by characteristics of the setting (Hannagan and Larimer 2010).

One clue affirming this hypothesis can be found in Susan Beck’s observations and interviews of men and women serving on municipal councils (2001). This study is especially useful for our purpose because these are people who attend public meetings about politics—and therefore the setting and the participants represent the types of settings and participants with which we are centrally concerned. Beck’s interviews with these council members suggest that while men and women share much in common, the women tend to be less comfortable in this setting than men. Women tend to respond more than men to others’ needs, and they “strive for collegiality” (Beck 2001). As one councilman said to Beck, “women … will avoid positions that will offend a group in town. They respond to feeling and people” (Beck 2001, 567). Other councilmen echoed this notion of women being more emotional, intuitive, and concerned for the feelings of others. While this notion may come from an attitude of denigration toward women, or from a stereotype about women’s emotionality, women may in fact be more tuned in to the emotions of the people around them, on average.

In addition, as we saw above, women tend to pursue more facilitative styles of interaction in legislative settings. This may indicate not only a preference for cooperation but also a deeper underlying orientation to connect with others. Women endorse or use a more facilitative leadership style, on average (Dodson and Carroll 1991; Flammang 1985; Kathlene 1994). Beck found that female city council members saw themselves as more oriented to listening rather than finding a “quick management solution” (2001, 58). Sue Thomas concludes, “[among] women … influence is used for responsiveness to colleagues and constituents rather than for personal gain” (Thomas 2005, 252). These conclusions are consistent with those from a broad array of studies of female leadership in lab and organizational settings (Eagly and Johnson 1990). As Eagly and Carli summarize, women are somewhat more likely than men to adopt a democratic style, even when they fill the same leadership position (2007, chapter 8). As a colleague of the female speaker of New York’s City Council said of her, “she has injected more democracy, with a little ‘d’, into the council; every single council member has a say in the budget” (Eagly and Carli 2007, 126).

A noteworthy illustration of women’s tendency to connect with others comes, again, from the work of Lyn Kathlene. She conducted a fascinating analysis of the words Colorado state legislators use during a scholarly interview to describe the resources they draw upon in formulating bills. She found that only women mentioned “citizens, community, country, district, parents, or world” and “city council, county commissioners, leaders, legislators, mayor/s”; while only men mentioned “advisers, clients, consultants, experts, industry” (Kathlene 2001). Women were more likely than men to see their key political actions as linked to their community, including their legislative or political community. During legislative debate, these legislators talked in ways that reflected gender differences in views about autonomy versus connectedness. Men tended to articulate the view that criminals are autonomous individuals making bad choices for which they should be held accountable; women tended to express the view that criminals are embedded in a structure that denies them opportunity, including unhealthy families, inadequate economic opportunity, or poor education (Kathlene 2001).²⁶ The differences are not merely the result by party. All these threads woven together create a pattern pointing toward women’s greater emotional and psychological interdependence with others. The small, narrow sample precludes generalization, but the results are suggestive of the hypothesis that women may tend to prioritize social bonds.

Suggestive evidence in this direction also comes from Beck’s interviews of suburban council members. A common refrain among her female, but not male, interviewees was that the politics surrounding them was “distasteful.” Partisanship bothered and even enraged some of the women. They viewed it as “backstabbing,” “political junk” (Beck 2001, 62). Again, this could be a preference for cooperation over conflict, but it could also stem from the desire to be part of a unified whole. What these women wanted above all, according to Beck, is to be part of a community of representatives. The “hardest thing” for these women was, in one woman’s words, “not working together” (2001, 63). By contrast, men’s most common complaint was against citizens voicing their views at council meetings (2001, 63). Men tended to resent citizen input; women tended to lament the lack of community. Put differently, on average, women may lean toward communitarianism rather than simply wishing to avoid conflict and live in peace.

Consistent with this finding, Karpowitz and Frost (2007) find that in public hearings about Walmart, women were significantly more likely than men to invoke the collective identity of the town and the connections among citizens when testifying before the town council. Instead of these communitarian appeals, men, by contrast, talked more about the merits of the legal arguments or about how Walmart might affect government revenues.

In sum, women tend to be more motivated than men by social ties and frequently do more to form and maintain connections to other people. On average, they may need and seek a heightened sense of community in their environment. They may respond more strongly to the absence of that sense of community. Situations in which people are not interacting cooperatively may trigger a sense of alienation or distance for women more than for men.

It stands to reason, then, that the deliberating group’s displays of regard for its members will affect women more than men, on average. Part of what discussion can do to encourage women’s participation is to signal to members that other members have an interdependent orientation. Discussions in which these expressions of social bonds are weak may expand the gender gap in participation; those with stronger positive ties, or warmer tone, may shrink it.

So far we have asked how women may differ from men, on average. We have developed three mechanisms or gender pathways that could explain why women may exercise less authority in discussion than men: because they are less confident, because they are more averse to conflict, and because they are more sensitive to social bonds. These three mechanisms are not mutually exclusive, but they are distinct enough that we can test each one.

However, the more central task of this book is to understand when differences between men and women mute. Put differently, what we really want to know is: what group-level conditions enlarge the gender gap in participation and influence? And what conditions shrink it, thereby helping women to achieve their potential for full citizenship?

GENDER COMPOSITION

When Women Interact with Men

Deliberation occurs in groups. This may seem to be an obvious fact, but it has received little systematic attention from studies of deliberation.²⁷ If we want to understand the gender gap in authority, we need to pay attention to ways that groups produce that gender gap and consider the possibility that group dynamics can affect women and men differently. By understanding how groups can create or reinforce gender gaps, we can also come to understand how groups could mute or erase those gaps.

To set the scene in broad strokes, we rely on the notion that gender is a characteristic not only of an individual, but also of a situation. Gender signals expectations about social interaction as much as it does about the characteristics of a person. That insight may be the most important development to come from the study of gender so far.²⁸ To be sure, we need to attend to the gender gap in the sources of participation and influence, which may lie in the fact that women are more likely than men to underestimate their competence, to alter their confidence in response to social feedback, to avoid asserting their opinions, to minimize conflict, and to prioritize their connection with those around them. But while it matters that women tend to have proclivities that differ from men’s, it matters even more that the effect of those gendered proclivities depends heavily on the setting. Our interest in particular lies with the ways that interaction within a discussion group may elicit gendered attitudes and behaviors, and how those gendered attitudes and behaviors produce a gender gap in participation and influence—in other words, in substantive and symbolic representation, in authority, standing, and power.²⁹ The way that groups are set up, and how they interact, can either facilitate or neutralize the detrimental effect of insecurity, reactivity, conformity, and conflict aversion on women’s speech. The composition of the group, and its procedures, can shape the norm of interaction, and the result is that gender roles may strongly determine individuals’ behavior, or conversely, be neutralized.

The first question to ask in our study of gender in groups, then, is who is in the room. Put differently, a key aspect of gender in groups is the relative number of men and women in the group. Scholarship on group composition suggests that gender composition has a variety of powerful effects on the group decision and on individual attitudes.³⁰ For example, gender composition affects judges’ decisions, and the level of respect in legislative debates—even when individual gender does not affect preferences, and even after controlling on ideology and other factors.³¹

Why would gender composition matter? One possible reason is the gender gap in perceived competence and in confidence. As we indicated, men still tend to be perceived as more competent in areas not clearly denoted as women’s domain, and that includes matters of common concern in general and public affairs specifically. And as we also noted, women not only tend to view themselves as less competent than men do, but they have less confidence in a variety of skills, and in formal situations, including public speaking. When women interact with men, they may view themselves as less qualified than the men in the group and be less motivated to assert their views. Consequently, when women discuss matters of common concern in mixed-gender groups, they may speak less, feel less confident, exercise less influence than men, and feel less free or able to express views or raise topics not articulated or shared by men, and the more men there are, the more this gender gap is likely to grow.³² The signal that gender composition sends about women’s status in the group may be subtle and indirect, yet powerful. Experimental studies show that men enjoy a higher status than women in discussions, unless the subject is commonly perceived to be a feminine one (Ridgeway and Smith-Lovin 1999). This would apply to discussions of most public affairs topics or matters of common concern to the group.

For example, Wood and Karten’s experiments show that when four-member, mixed-gender groups are asked to reach a unanimous decision in a hypothetical ethical dilemma, men are more likely than women to engage in leadership behaviors such as offering information, making suggestions, and giving opinions, and to be perceived as competent. Moreover, in another experiment they show that this gender gap goes away when the investigators supply the same (false) competence feedback to the men and the women before discussion (Wood and Karten 1986). The overall pattern suggests that the reason men exercise more authority and are perceived as more authoritative is that they are viewed, and view themselves, as relatively more competent than women. Thus when women interact with men on such topics, women’s relatively low status may translate into inequality in participation, representation, and influence. In mixed-gender situations, the group thus becomes a site for the enactment, and reinforcement, of gender role expectations that posit that women have less authority than men.

An example of a study that underscores the effects on women when they interact with men is the Israeli engineering study of Gneezy, Niederle, and Rustichini (2003). Gneezy and his colleagues assigned men and women at a highly selective engineering school to solve maze problems. We noted above that the more competition, the worse women did and the better men did. However, competition is not the end of the story. Our point here is that in addition, the effect of competition was highest when women had to compete against men. In fact, when women competed exclusively against other women, they solved more mazes than they did without competition of any kind. So women may not do worse in conditions of competition per se. Rather, they may react negatively to competition when they compete with men. Because men are viewed as more competent engineers, having to compete with men depresses women’s performance.³³ Similarly, in discussion with men on subjects other than those considered women’s special purview, women are likely to view their competence as inferior to men’s, and make fewer attempts to express their views or influence the group.³⁴

Not only are women likely to take the group composition as a signal about their relative competence and authority, those around them are also likely to alter their behavior accordingly. The more men in the group, the higher the number of individuals who are likely to signal to women that their contributions are less adequate. A recent review of psychology studies of leadership found that stereotypes that associate leadership with masculinity are more commonly held by men than by women; that is, men are especially likely to view leadership as a masculine role and to see men as more fit to lead (Koenig et al. 2011). Along these lines, men are also more likely than women to negatively evaluate women who engage in leadership acts even when these acts are identical to those undertaken by men (Eagly, Makhijani, and Klonsky 1992). Similarly, not only are men less likely to choose women than to choose men for political discussion,³⁵ men are also less inclined than women to attribute influence to authoritative women when they speak in public. Specifically, men in churches whose clergy is a woman rate the political influence of the clergy on the members much lower than do female members of those churches, and underestimate the amount of political speech issued by the clergy member (Djupe, Sokhey, and Gilbert 2007).

These patterns matter to the level of influence that women can exercise when speaking with many men. They demonstrate that men are likely to infer that women are less competent than they are in discussions of matters not deemed feminine, and as a consequence of that evaluation accept less influence from women than they do from men. And that in turn means that when women are interacting with many men, they will carry less influence than when interacting with many women. A set of studies on gendered double standards and influence find that men tend to devalue women’s competence, and as a consequence are less likely to accept influence from them. Furthermore, women do not devalue men’s competence, and consequently they are likely to accept influence from them. In these studies, pairs first individually performed a task that is not viewed as feminine. They were then informed either that they outperformed or underperformed their partner. They were then observed interacting about disagreements prearranged by the investigators, and the level of influence from one to another is recorded. The findings show that men are less influenced by a female partner than women are by a male partner, unless she outperformed him in the first round. A woman accepts influence from the male partner even when he underperforms her, while a man accepts influence from his female partner only if she outperforms him. Men are more likely than women to expect less of women’s ability and consequently tend to accept less influence from women than from men.³⁶

These results suggest that when men are women’s predominant conversation partners, women will receive signals that they have less to contribute than the men in the group. They will be expected to behave in less assertive and more accommodating ways than they would if surrounded by many women. As one respondent told interviewers in Conover and colleagues’ study of public discussions, “the men are talking, and the ladies and, you know, the wives or whatever, and we are almost spoken down to—because they know better than us! So I think we’ve had to keep quiet, you know [nervous laughter]” (Conover, Searing, and Crewe 2002, 56). The more men present, the greater the number of members likely to hold stereotypes that regard authority as men’s purview and who thus may act in ways that send negative signals to women who engage in acts of authority.³⁷ Even when the signals are subtle, women may pick up on these expectations to remain relatively quiescent and unassertive during discussion with men.

Finally, when women interact with many men, they are also less likely to engage in assertive behaviors than they do when they interact with women. For example, women who have a personality that seeks dominance take dominant positions primarily with respect to their female conversation partners. Women who score high on dominance measures attempt to assert dominance over women who scored low on dominance, but not over men who scored low in dominance.³⁸

All this is to say that interactions between men and women tend to be characterized by unequal authority. We can tell that authority is at work here, because when status is equalized between the sexes, their inequality disappears. For example, in studies of the conversations of intimate heterosexual couples, members of couples characterized by an equal relationship exhibit equal talk times (Kollock, Blumstein, and Schwartz 1985). Similarly, women who fill clearly legitimized leadership roles, such as managers in an organization, engage in leadership acts as often as men do in these roles (Eagly and Johnson 1990). By default, gender is entangled with authority; when authority is equalized, gender differences tend to deflate and even disappear.

In mixed-gender discussion of matters of common concern, women may defer to men’s perceived expertise, and the result may be less talk, less talk of women’s distinctive concerns, less talk about women’s preferences, less affirmation of women’s speech, less perceived influence over the discussion, and more adverse consequences for women’s sense of efficacy.

The Minority Status Hypothesis

The foregoing leads to what we call the “minority status” hypothesis: numerical minority status affects women’s status in the group (see table 3.1). That status in turn affects their participation in discussion, their substantive representation—how much they speak to their particular concerns and preferences—and their symbolic representation—whether they are viewed as full citizens with equal authority to govern. Women have the most standing in groups with many women, and they face the greatest disadvantage when they are heavily outnumbered, especially when one woman is surrounded by men. In general, the more sparse women are, the lower their status, and the wider the gender gap in participation and representation.³⁹ Support for this notion comes from various studies. For example, Johnson and Schulman (1989) found that while both men’s and women’s influence is rated lower when they are in a numerical minority, women incur a greater disadvantage.⁴⁰ A study of small groups’ evaluation of candidates for promotion at a utility company found similar effects on evaluations of competence (Schmitt and Hill 1977).⁴¹

This effect may be due to the inferences people draw about gender and competence when women are scarce. Women in male-dominated work groups experience more sex stereotyping than women in female-dominated groups (Konrad, Winter, and Gutek 1992, 131). In one study, students working on Masters of Business Administration (MBA) degrees were more likely to attribute stereotypically feminine traits to women in an applicant pool for a hypothetical job, and to rate them more negatively, when the pool included fewer than three women out of eight than when the pool included a larger proportion of women (Heilman 1980). In addition, in an analysis of 486 jobs across several firms, the percentage of women in blue-collar and clerical jobs in a given firm was positively related to supervisors’ ratings of women’s performance, even after accounting for women’s education and ability (Sackett et al. 1991, 265). These studies imply that women’s disadvantage when they are a numerical minority is at least partially caused by the stereotypes primed by their minority status per se. When people see few women, they may implicitly assume that the task at hand is not going to be well performed by women. That is, the number of women may act as a cue to gender stereotypes about the kinds of actions that women are well suited to undertake.

Table 3.1: Main Hypotheses about Gender Composition and Decision Rules of Small Groups

Hypothesis	Explanation
Minority Status Hypothesis	The average woman in the numerical minority occupies a lower status and hence participates and influences less than the average woman in a numerical majority. Consequently, the gender gap in the group shrinks as women’s numbers increase. Men may be negatively affected by being a minority gender, but not as much as women.
Enclave Hypothesis	The average woman’s participation and influence is greatest in all-female groups, where women are less constrained by the gender role expectation to avoid authoritative behavior, and which build women’s confidence and employ cooperative norms. There are no clear expectations for male enclaves.
Interaction Hypothesis	Under majority rule, the average woman participates and influences more as women’s numbers increase, shrinking the gender gap in the group. Men may be negatively affected by being a minority gender, but not as much as women. Under this rule, the minority status hypothesis is accurate.
	Under unanimous rule, the group develops an inclusive interaction norm that elevates the participation and influence of both gender minorities. When women are few, that norm increases the average woman’s participation and influence. Thus the average minority woman participates and influences considerably more under unanimous than majority rule. But the inclusive norm also raises the average minority man’s participation and influence. Therefore, under unanimous rule, the average woman’s participation and influence remains the same, or even decreases, as women’s numbers increase, enlarging the gender gap in the group. That is, the effects of women’s higher numbers are neutralized, or even exceeded, by the boost minority men received from being the gender minority under consensus norms of interaction.

Once these gender stereotypes are cued, they may implicitly affect women’s actions. Experiments conducted on “stereotype threat” demonstrate that even seemingly minor signals about women’s relative competence compared to men can significantly affect women’s performance on tasks viewed as more masculine. For example, one set of Michigan undergraduates was told that there were usually no gender differences on a difficult math test they were about to take. Another set was told that the test usually produced gender differences. Women scored less than half as well in the second than in the first group. Men moved in the opposite direction, gaining a performance boost from the gender signal (Spencer, Steele, and Quinn 1999). The idea behind findings such as these is that others’ expectations reside in the person’s mind in the form of stereotypes and associations between who they are and what they are capable of. People learn these associations over the course of countless small, often subtle interactions with their environment. Men and women know about these and are affected even when they are unaware of that effect and even when they have not internalized the stereotypes, indeed, even when the stereotypes are rejected (Aronson et al. 1999). When these expectations are made salient, or “primed”—for example, by the scarcity of members of the person’s social category—the person’s confidence in their competence drops. And the less confidence individuals feel, the less likely they are to attempt the action, to execute it well, and to feel good about their attempt. This is the theory of stereotype threat, and it has been validated over many settings (Steele and Aronson 1995). When women are under conditions of stereotype threat, their performance relative to men on masculine-typed tasks suffers (Kray, Thompson, and Galinsky 2001).

This effect of women’s relative numbers is likely to be all the more powerful when people interact. Not only will women observe many men around them and draw the implicit conclusion that they are not suited to the task at hand. The men are likely to act confidently, exacerbating women’s drop in self-confidence. The journalist Sara Laschever recalls that after having worked as an executive, she took a temporary job as a store clerk where she was required to perform mental arithmetic calculations. Her (male) boss, however, did not know of her competence and believed that he was much better at the task. He exhibited high competence, and he was, well, a man. Although she had always done well at math, she found herself making mistakes and feeling unsure of her ability (Babcock and Laschever 2003, 80). The economist Linda Babcock recounts a similar story. While serving as an interim dean, Babcock made a presentation at an important meeting with university officers. She had always felt well respected and valued by these colleagues, but in this meeting, she found herself the only woman. She grew “uncharacteristically nervous,” and “when it was her turn to speak” she found herself “petrified” (81). Her self-confidence dropped and negatively affected her performance. She spoke less well than she believed herself capable. The effects of others’ expectations work through women’s own self-confidence.

The key point is that who is in the room can have a powerful effect on what women think, feel, and ultimately do during the group’s discussion. When the situation cues women’s sense that they may not belong, and that they may not be as competent as men, their lowered self-confidence may thus prompt them to underperform and perhaps to withdraw. They may lessen their participation in discussion, raise issues only if they believe the others are likely to view those issues as relevant, express preferences only if they anticipate that they are shared by others in the group, and respond strongly to negative signals by other deliberators. As the research on stereotype threat shows, and in line with the general literature on stereotypes, this process may not be conscious and may occur despite a woman’s rejection of gender stereotypes.

There is reason to expect women to participate less and to achieve lower levels of substantive representation in groups where women are a minority. In these settings, women’s low numbers, and the presence of active, confident male participants, will place women in a lower status relative to men. Women will perceive themselves to have less authority than men, and consequently they will speak less, speak less to their distinctive concerns, articulate their preferences less often, and carry less influence over the group.

This is what we mean by opportunity and the motivation to participate. Women need discussion settings that do not cue their relative lack of confidence, relative lack of assertion, and sense of relative incompetence in comparison with men. Even when the signals are not overtly hostile, the absence of positive reassurance can act as a negative cue. Society defines women as members of a lower-status category; when women are a minority, that status becomes salient, and the consequences follow predictably.

Gendered Norms of Interaction

The prediction of the “minority-status” hypothesis is seconded by another literature. Women may participate less and carry less authority in discussions with many men than in discussions with many women not only because of a gender difference in perceived expertise and confidence, but also because gendered norms of interaction vary with gender composition and facilitate or hinder women’s participation. Specifically, predominantly female groups tend to adopt more inclusive and cooperative norms of interaction, while predominantly male groups tend to accent individual agency. Women tend to be deterred by the latter and encouraged by the former.

We have already discussed the fact that women are socialized to interact in more expressive, cooperative, and warm ways, but our point here is that small group interaction amplifies this tendency. The more cooperative and warm individuals are present, the more the norm tilts in that direction. In settings with many men, the norm of interaction tends to take on more stereotypically masculine characteristics of individual assertion, agency, competition, and dominance; in settings with many women, people tend to interact in a more stereotypically feminine style that emphasizes cooperation, intimacy, and the inclusion of all participants.⁴² This, too, is what we mean when we say that gender is a feature of the context, not just of individuals. People take their cues about gendered behavior from the gender of those around them. When surrounded by many men, both men and women are likely to shift their own behavior in a more masculine direction, emphasizing individual agency and assertion. It’s as if people note the gender of those around them and conclude that the setting calls for acting like a member of that gender. Composition signals the social expectation of conforming to the behavioral style of the numerically prevalent gender.

A high-level female executive in a male-dominated industry illustrates how individuals shift in response to the perceived masculine norm: “I learned that you had to … put on a more serious demeanor, to establish credibility more quickly … you stop trying to be warm, wonderful, and nice” (Eagly and Carli 2007, 124). Female leaders tend to adopt a stereotypically masculine style in predominantly male settings, pursuing task-oriented rather than democratic or nurturing leadership to the same extent as the men in these settings, and adopting a more tough and domineering manner with subordinates (Eagly and Carli 2007, 126). As we noted earlier, this style tends to trigger social dislike, and yet despite the social cost, women in highly masculine settings, and occupying a role with heavily masculine expectations, tend to find it necessary to exercise power in the ways defined as masculine. Examples of this phenomenon are well known: Indira Gandhi, Golda Meir, and Margaret Thatcher.⁴³ In less male settings, however, female leaders tend to behave in ways that are “caring, supportive and considerate” (Eagly and Carli 2007, 130). When surrounded by many women, people expect themselves and others to behave in more stereotypically feminine fashion—to interact in ways that are more warm, expressive, and cooperative.

Evidence that the gender composition of the setting affects gendered norms and behavior comes from a variety of studies. For example, levels of self-disclosure during group discussion increase as the number of women in the group increases (Dindia and Allen 1992). For another example, men in all-male university departments express the least cohesion, and men in predominantly female departments express the most (Bird and Wharton 1996, 109; Reskin, McBrier, and Kmec 1999, 347). All-male groups establish hierarchy more quickly than all-female groups, and when women are the majority, the group develops less hierarchy (Berdahl and Anderson 2005; Mast 2001). Similarly, when a confederate acts in a dominant, aggressive manner during group discussion, all-male groups are more likely to gang up on the aggressor with dominance behavior of their own, relative to all-female groups (Ridgeway and Diekema 1989). A study of interruption patterns in simulated work groups found that both men and women issued more hostile interruptions as the proportion of men increased (Karakowsky, McBey, and Miller 2004).⁴⁴

These findings are consistent with the culture theory of gender. A key piece of this theory is that gendered cultural scripts of behavior are triggered by gender composition. That is, a gendered cultural convention is acted out when the individual interacts with others who follow the same cultural convention. An individual surrounded by many women will adopt the conventions of women in the same way that an individual surrounded by many Americans will adopt the conventions of Americans. The more homogenous the gender of the small group, the more that every member will conform to the social conventions of the predominant gender. An all-female group signals implicitly that members will interact according to a feminine style; the same holds for all-male groups and their cue to use a masculine style (Maltz and Borker 1982; Tannen 1990; Thorne and Luria 1986). These gendered subcultures are said to be learned early; some of the same gender composition differences scholars find in adult groups show up in girls’ and boys’ play patterns (Maccoby 1998). As we mentioned earlier, for example, girls’ playgroups tend to avoid conflict within the group and to reach agreement more than boys’ groups do (Miller, Danaher, and Forbes 1986). Thus the theory argues that girls and boys are socialized to different gendered cultures of interaction and carry these implicit scripts of behavior with them into adulthood. This theory can explain not only why gender composition shapes gendered norms of interaction but also highlights the independent and differential effect of those norms on women versus men.

The theory of gendered subcultures is consistent with the findings of several studies we reviewed above. For example, it predicts women’s proclivity to dislike conflict and their sensitivity to affiliative social ties, and that women respond to settings that highlight conflict or cold relations by reducing their participation. The prediction is grounded in the notion that women do so out of a sense of cultural alienation. Other settings highlight cooperation and warm interactions, and these may encourage women to take full part. Again, this positive boost is grounded in women’s cultural milieu, which creates a sense of comfort and cultural familiarity with settings where individuals work to create rapport.

One example of how gendered norms play out is in social science classrooms. A study of Harvard undergraduates identified a style of discussion called “gov talk” (Margolis 1992). “Gov talk” is depersonalized, focuses on abstractions, and is rationalistic, according to the author. A number of women interviewed in the study claimed that this style made them uncomfortable, depressed their frequency of speaking during class discussion, and caused them to be “unwilling to challenge others’ views in public” (Hansen 1997). These suggestive findings are consistent with the findings we reviewed in chapter 2, which show that women participate less than men in activities that express clear opinions despite their balanced numbers in many educational and civic settings. Other evidence we reviewed earlier also fits this theory. For example, the theory is fully consistent with Kathlene’s (1994) findings about predominantly male legislative committees exhibiting a competitive, aggressive communication behavior that tends to inhibit women’s participation more than it does men’s. The theory predicts that such behavior will not only be found in university classrooms or formal legislative committees; rather, in a broad swath of settings with more masculine interaction norms, women will participate and influence less, and speak less about what’s on their mind.

Men may also be affected by gendered norms of interaction around them, although to a lesser extent.⁴⁵ As women’s status in the group grows, men’s may lessen. Men may participate less and be less likely to raise issues of distinctive concerns to men. The evidence, however, is sparse, and the effects tend to be small. Some indirect evidence comes from studies of the effect of women’s candidacies on citizens. Susan Hansen found that during the 1992 presidential election year, men were somewhat less likely to engage in political proselytizing—attempts to persuade others to vote for one’s preferred candidate—if women were on the ballot (Hansen 1992). As discussed in chapter 2, the effect of female candidates is positive on women; as it turns out, men experience a slight drop in engagement as women rise in it. These findings hold after controls for the person’s partisanship, age, party, and issue positions, so they seem to be genuine gender effects rather than spurious consequences (Hansen 1997).

The gendered culture theory argues that mixed-gender settings are characterized by a norm that deviates from the one to which women were acculturated, and that this bad fit depresses women’s sense of belonging to the group. Women may thus experience a greater sense of comfort in predominantly female settings with their more stereotypically feminine norms of interaction and feel ill at ease in settings with predominantly male norms. There are far less clear expectations for the effects of gender composition and of masculine or feminine norms on men. Thus the literature on gendered norms seconds the “minority-status” hypothesis—women will participate less, and engage in other forms of substantive and symbolic representation less, in predominantly male groups, and increase their involvement as their proportion increases.

The Enclave Hypothesis

Existing studies of group gender dynamics also support an “enclave” hypothesis: women flourish in all-female settings. This follows from the logic of the work on minority status. Because women are disadvantaged in political discussions with men, they may do best without any interactions with men and may benefit most from their own discussion space. Thus women’s talk time, their references to women’s issues, and their perceived and actual influence will be higher in all-female than in mixed-gender deliberating groups (see table 3.1).⁴⁶

Support for the enclave hypothesis emerges from a variety of sources. For example, the philosopher Habermas might approve of female enclaves. As we noted in chapter 2, feminist theorists criticize Habermas for paying inadequate attention to ways that gender inequality undermines women in deliberation (Fraser 1990). But Habermas was quite taken with what he viewed as protected social spaces for people, located away from the influence of political power—namely, the church and the government. For Habermas, civil society is the “nervous system” of democracy for this reason (McCarthy 1994). The salons of the French enlightenment, and the coffeehouses of eighteenth-century Britain, served for Habermas as a key historical lesson about the need in democracy for deliberative spaces that the authorities cannot infiltrate. In these spaces, citizens could develop a public opinion independent enough of entities that hold disproportionate power to be able to hold these powers accountable.⁴⁷

Nancy Fraser has specifically argued that women have long formed their own civil society and recommended these protected enclaves as a form of “subaltern counterpublics.” The idea is similar to Habermas’s notion that people with less power need to interact with each other as a way to form opinions uncontaminated by coercion, or worse still, by the more implicit but insidious influence of others who control the hegemonic ideology of the day.

Karpowitz, Raphael, and Hammond (2009) take on the notion of enclaves directly, arguing that the goals of deliberative democracy are well served by protected spaces in which homogenous groups—especially those who have been traditionally disempowered—can congregate, talk among themselves, and explore the issues of the day from their own unique perspectives. Enclaves are not, as those who worry about group polarization often assume, only groups with common views; they should also be understood as groups that might share a structural location with respect to an issue or a common predeliberation identity. In fact, one of the values of enclave deliberation is that it allows groups with such a common identity not just to practice civic skills but also to more fully understand their similarities and their differences. By allowing such exploration of the diversity of views within groups that are homogeneous in other ways, enclave deliberation “can thus serve the larger cause of a fully inclusive public discourse by giving disempowered or marginalized groups an opportunity to develop their own unique perspectives and arguments, which might otherwise be overlooked or ignored” (Karpowitz, Raphael, and Hammond 2009, 582). This can be the first step in more effective and successful engagement in mixed settings as part of the larger deliberative landscape.

Do American Women Congregate in Enclaves?

Of course, the benefits of enclaves are only possible if women do, in fact, congregate among themselves. And on that score, women in the United States have a rich tradition of enclave endeavors; in part because of their formal exclusion from some aspects of civic life, women joined all-female associations through much of American history. In Skocpol’s words, “voluntary associations have always rivaled voting as pathways Americans follow into community and public affairs” (Skocpol 1999, 462). A key feature of these groups is that their members interacted. As Theda Skocpol describes it, their watchword was “interact or die” (Skocpol 1999, 491). The point we want to emphasize is that in these all-female settings women interacted only with each other. Here they had the chance to develop their political skills and aptitudes. They were trained to lead, whether through a general sense of capacity and support from fellow women or by learning practical acts—“giving a speech, running a meeting, keeping the books” (Burns, Schlozman, and Verba 2001, 73). The first women to become prominent political leaders were members of all-female associations.⁴⁸ And Skocpol (1992) recounts how women used women’s associations to achieve meaningful policy change, including the rise of the Children’s Bureau, mother’s pensions, and other elements of the early welfare state. Even today, women’s associations have a great deal to do with whether women in their area run for local office (Crowder-Meyer 2010).⁴⁹

As women mobilized for gender equality in the 1970s and 1980s, they did so largely in all-female groups. These included the National Organization for Women, which was based in local chapters where women met with their neighbors and members of their community (Skocpol 1999, 468), and other new organizations, some of which also included interaction among female members. They joined women’s groups with a long-established tradition of local meetings, including the YWCA, League of Women Voters, the American Association of University Women, the General Federation of Women’s Clubs (GFWC), and the National Congress of Parents and Teachers (Skocpol 1999, 465–66). The GFWC alone had nearly a million members and over 15,000 clubs at its peak (Skocpol 1999, 479).⁵⁰ While old civic groups declined, a trend well documented—and well lamented—by Robert Putnam in Bowling Alone, new groups oriented to advancing disadvantaged people surged (Skocpol 1999, 468–72). Women’s rights groups, composed almost entirely of women, were at the crest of the surge. One scholar found that while various racial, ethnic, and women’s associations proliferated in that period, women’s groups did so exponentially.⁵¹

Women still belong to all-female or predominantly female groups. Evidence on this comes from the 1990 Citizen Participation Study, the backbone of PROPA. Burns, Schlozman, and Verba asked their respondents to which organizations they belonged, defined as being either a member or a contributor (2001).⁵² Then they asked respondents to indicate which one of these organizations is the “most important” to them. Finally, they asked about the gender composition of this “most important” organization. Of women, 17% report that their most important organization is all female, another 17% report belonging to a mostly female group, 60% report that their group is “mixed,” and only 4% report “few” women (Burns, Schlozman, and Verba 2001, 76).⁵³ While women are not in all-female groups as a rule, there are certainly still settings where women do interact exclusively with each other.

Burns and her colleagues report that many men also belong to a “most important” group composed entirely or predominantly of men. In fact, men are more likely than women to belong to a “most important” group composed “mostly” of the same sex. The percentages of men reporting all male, mostly male, mixed gender, and few men are, respectively: 13, 26, 56, and 2%. Men, too, have opportunities to interact exclusively with other men.⁵⁴

The workplace also continues to be a place where some people interact in predominantly same-sex groups. A national survey of American organizations found that 11% of businesses had over 90% male employees, and 7% of businesses were more than 90% female (Reskin, McBrier, and Kmec 1999, citing Kalleberg et al. 1996).⁵⁵ Approximately 90% of businesses in the United States employ fewer than twenty employees, a number small enough to generate interaction among the employees.⁵⁶ So in nearly 20% of businesses, people are interacting almost exclusively with members of the same sex.

These findings underscore an important point for our purpose—there is considerable variety in the gender compositions of American organizations, from work groups to political boards and councils to voluntary organizations of various kinds. While a majority of Americans are in mixed-gender groups, a sizable minority of Americans are in groups where one gender or the other predominates. This means that it makes sense to ask, as we do, what is the effect of this variety on its participants?

What Happens in Enclaves?

What happens to women who associate with women?⁵⁷ Burns, Schlozman, and Verba (2001) can tell us a great deal, at least, about what people report is happening to them in their civic associations. They asked their respondents a series of questions about their activities in their “most important” organization. In gender-mixed organizations, women are less likely than men to report activity.

The gender gap in mixed-gender groups is largest on the activities that are most relevant to leadership of discussion: whether the person has written a letter, the number of times the person has been asked for their opinion, the number of times the person has expressed an opinion at a meeting, the number of times the person went to a decision-making meeting. And the biggest gender gap of all obtains on whether the person made a speech (men exceed women by 12 percentage points).⁵⁸ In other words, where women trail men the most in gender-mixed groups is on the act that most requires a confident, public statement of views—in other words, the most leader-like act on the list—making a speech.

By contrast, “plan a meeting” produces only the tiniest of gender gaps. A general indifference to meetings is not the reason why women are less likely to express an opinion at a meeting, attend meetings where decisions are made, or make a speech at a meeting. Women are not averse to getting involved with meetings, but they tend to engage in meetings more as support staff than as influential members or leaders. Women might send out mailings announcing a meeting, or figure out where to seat the participants, or provide other logistical aid. Support fits women’s traditional gender role much more than does leadership. And that feminine gender role may get in the way of women’s full exercise of voice in the life of their communities.

But this gender gap reveals itself only when we compare women and men in mixed-gender organizations. In same-sex associations, these gender differences disappear or even reverse.⁵⁹ The consistency across the various activities is striking; there is not a single activity that sustains women’s disadvantage to men in same-sex groups. The gender gap disappears even though men increase their activities in same-sex groups. That is, women gain so much from same-sex groups that they close the gap with men despite men’s own gains from their same-sex groups.⁶⁰

Even more telling of the empowerment of enclaves, the activities on which they elevate women’s participation the most are those most relevant to authority. The activities with the five biggest increases in women’s percentages are as follows, with the percentage point increase from mixed to all-female groups noted in parenthesis:

•  Went to a meeting where respondent took part in making decisions (31%)

•  Opinion has been asked (19%)

•  Expressed opinion at meeting (18%)

•  Wrote a letter (16%)

•  Made a speech (15%)

•  Served on board or as officer (15%)

The magnitude of these enclave effects is remarkable. A 30+ percentage point effect, which is what we see for attending decision-making meetings, is nearly impossible to unearth in the study of participation.⁶¹ But that effect is specific. The massive shifts occur mostly in women’s involvement in decisions made for the collective. One has to think of oneself as entitled to and responsible for contributing to and influencing decisions in order to attend meetings where decisions are made. Making decisions is at the heart of authority.⁶²

In sum, as PROPA describes it, “organizations of women provide the kinds of experiences that have been attributed to them: providing opportunities for leadership, facilitating the exercise of voice in organizational matters and the development of civic skills, and generating requests for political activity” (Burns, Schlozman, and Verba 2001, 230). But in our view, what is significant about the findings is less that they locate the source of women’s civic skills, but more that they point toward a particular mechanism—authority. Women need more than the knowledge of how to plan a meeting; they need, far more, the sense that they are entitled to power. The results from that study’s data confirm the importance of looking at all-female groups for the development of authority.

However, as important as these findings are for their implications, they require some caveats. They emerge from a research design that serves some purposes quite well, but which leaves some doubts for our purposes, as we will explain in greater detail in chapter 4. For now we note that the data come from self-reports rather than objective measures of gender composition and of the person’s experiences in their group; that a person who ends up in an enclave may differ in important respects from one who does not, and this initial difference could explain the apparent effects of enclaves; and that enclaves may differ from mixed groups in any number of ways that could account for the differences we observed.

The problem of self-reported experiences, and of spurious differences between enclaves and others, can be addressed with controlled experiments. Existing experimental studies provide support for the observational findings from surveys. In general, single-sex groups generate larger gender differences than do mixed-sex groups (Aries 1996). Similarly, in their study of how much individuals disclose about themselves in group settings, Dindia and Allen use a meta-analytic approach to show that all-female groups display high levels of self-disclosure, all-male groups very low levels, and mixed groups fall in between these extremes (1992). Studies of negotiation also show that in all-female dyads, negotiators share more information, do so earlier in the exchange, attend more to the needs of the other party, and the agreements they reach are more likely to serve the common good, relative to mixed-gender or all-male pairs (Babcock and Laschever 2003, 168–72). A similar pattern emerges in studies of dominance behavior.⁶³ Women in enclaves are more likely than women in mixed groups to initiate interaction and to be addressed (Aries 1976, 15).⁶⁴ Women are more likely to express emotion and to read other people’s emotion accurately from nonverbal signals, especially when reading women’s signals. The implication of this is that women will be more tuned in to other group members in a mutually reinforcing cycle of emotional sensitivity, leading to more group solidarity.⁶⁵ While these studies have their own methodological problems, detailed in the next chapter, they do offer enough support for the enclave hypothesis that it is worth testing.

The enclave hypothesis is worth further investigation. There are enough all-female and all-male groups in American public life that they are worth studying. Existing research points to the possible qualitative differences between mixed-gender and all-female groups. These differences suggest that all-female groups may be a place of particular empowerment for women. The existing studies leave methodological gaps, but we proceed with their working hypothesis: levels of women’s participation and representation rise from mixed-gender groups to all-female enclaves.

In summary, we will test several implications of the notion that women’s numbers matter. One is the “minority status hypothesis,” that women in a numerical minority occupy a lower status. The other is the “enclave hypothesis,” that women in all-female groups are especially empowered. We wish to see if these variables affect the level of women’s participation and representation in group discussion.

DECISION RULE

Gender role theory, including the minority status hypothesis and the enclave hypothesis, dominates the literature on gender and discussion, but we view it as incomplete. Specifically, the proportion of women in a group is not the only important factor that affects gendered patterns of speech participation. Institutions—by which we mean the rules and procedures that organize group functioning—can eliminate the disadvantages of low numbers; similarly, they can block the power of high numbers. Among the neglected procedural or institutional factors that shape group norms is the group’s decision rule.

A rule that governs the group can determine far more than the substance of its decision. It can set in motion a set of social scripts, cancel individual habits, and produce particular styles of interaction. In other words, rules can help to shape a group’s norms and dynamics, and as we have argued, these norms in turn can elevate or depress the authority of group members.

A rule is a relatively small intervention that carries large potential effects. Groups can often have a say over their own procedures. Gastil claims that “most small groups have considerable leeway in establishing at least some—if not all—of their procedures” (2010, 101). It can be far easier to adopt than a dramatic elevation in the physical presence of an underrepresented group. It can be adopted by the group itself, or imposed by those empowered to design the discussion. It is a reform that can be imported into many different settings, is easy to understand, and does not prompt contentious debates about quotas or difficult controversies about the appropriate remedies to the low status of a social group. It is easy to implement and far-reaching in its benefits.⁶⁶

No rule is a panacea. Any given rule carries advantages and disadvantages, as we elaborate in the conclusion chapter. In fact, we leverage the contingent effect of a rule to spell out when a rule will benefit women, and when it will further erode their standing. Despite the caveats just noted, our point stands: some rules are likely to generate a process that can build the authority of women.

The two most common rules are majority rule and unanimous rule. Majority rule is commonly used in formal settings of politics. Unanimous rule has been investigated in recent years primarily in the context of jury decision making and groups with close bonds.⁶⁷ But unanimous rule and its variants are common in various settings, including formal institutions with a great deal of power, such as the UN Security Council and the US Courts of Appeals.⁶⁸

Unanimous rule holds out a significant potential to elevate the authority of women, because it can create group norms that enhance consensus and inclusion. Since the minority faces difficulty in persuading the majority in small groups (Moscovici 1980; 1985), a rule that includes minorities can substantially alter group dynamics. A seminal experimental study of mock juries reports that people shift their views during discussion more under unanimous than majority rule (Hastie, Penrod, and Pennington 1983). Such shifts are testament to the strength of the consensus norm that a rule can produce. The norm of cooperation can be seen in Nemeth’s finding that mock juries instructed to use unanimous rule produced more expressions of agreement than those instructed to use majority rule (1977).⁶⁹ Unanimous rule also leads to a fuller sharing of information by deliberators (Mathis 2011). Perhaps because it creates a dynamic of mutual information exchange, engagement, and inclusion, unanimous rule can increase a sense that the decision was legitimate and appropriate and renders it more widely and deeply acceptable.⁷⁰

That satisfaction is all the more telling when it extends to the preference minority, and in fact, preference minorities are also more satisfied with the group decision under unanimous than majority rule (Kerr et al. 1976). Under some conditions, groups distribute resources less equitably and form factions more often under majority than unanimous rule (Thompson, Mannix, and Bazerman 1988). As further indication that the process by which people interact shapes their views of the group, group consensus generated through talk can lead to increased cooperative behavior (Bouas and Komorita 1996). Thus unanimous rule leads to consensus-oriented norms of inclusion. These norms can protect numerical minorities beyond the direct veto power of the rule itself (Mendelberg 2002).

To be sure, under conditions of grave conflict, a group governed by the requirement to reach agreement may end up pressuring the minority.⁷¹ The requirement to reach agreement may end up cultivating a norm that suppresses disagreement.⁷² In that sense, unanimous rule does not benefit the numerical minority under all conditions.⁷³

Regardless of whether unanimity empowers the preference minority or muzzles it, no one has considered whether the protective effect that the rule can carry for preference minorities holds for social identity minorities. Even as scholars devote attention to the effects of rules and group processes on preference minorities, we know almost nothing about the effects on social identity minorities. We argue that unanimous rule, and the consensus process it tends to prompt, elevates the participation and influence of social identity groups when they are few. The most important claim we make is that norms of inclusion and cooperation tend to benefit women when they are the minority.

There are several reasons to expect that what unanimous rule does for preference minorities, it will do for gender minorities. In either case, the rule places an emphasis on inclusion and cooperation. As studies reviewed above show, the norm that unanimous rule creates is the expectation that everyone should be included in the decision making. As Mansbridge explains, “a consensual rule can actually create unity” (1983, 256). While unanimous rule may also pressure minorities to go along with the group’s central tendency at the end of the day, nevertheless, unanimous rule produces the expectation that each voice should be heard.⁷⁴ This may help those in the minority by elevating their level of participation—and influence. As Gastil hypothesizes, “consensus assumes that the minority viewpoint is crucial, so members may go out of their way to draw out quieter group members. Listening may also be enhanced, since consensus relies upon members understanding and considering what each other says. Without such listening it becomes far more difficult to arrive at a decision acceptable to all group members” (1993, 52).

Unanimous rule also signals that the group should orient to its members’ commonalities, and may create the expectation that the group is a social unit, and that decisions should be based on equal respect (Mansbridge 1983, 14). As Gastil argues, “it is more likely that consensus groups will direct energy toward maintaining a healthy relational atmosphere” (1993, 52). Under unanimity, no voice can be overlooked because every vote is pivotal. And we argue that this veto power generates an overarching emphasis on common bonds, strengthening the identification with the group as a social category and the view of the members as brethren. As Aristotle put it, “unanimity [homonoia] … seems akin to friendship” (quoted in Mansbridge 1983, 14), and as Mansbridge argues, “the rule of consensus seems not only to reflect empathy but to create it” (1983, 256).

Conversely, majority rule signals that conflict is acceptable and that some perspectives may not be included in the group’s final decision. Majority rule sets a norm whereby decisions are based on a contest of interests, and on numerical power. With this norm, majorities are less likely to indicate inclusiveness, and minorities are less likely to assume that their voices matter and that they should speak. As Gastil puts it, under majority rule, “majorities have no short term need to hear minority opinions” (2010, 99). Under majority rule the power of numbers matters most, and minorities are ultimately at a disadvantage.

The packaging of majority rule with other adversarial aspects of the polity is most clearly seen in Mansbridge’s foundational work, Beyond Adversary Democracy, which also draws a clear contrast between the norms of unanimity and majority rule.⁷⁵ “Majority rule” is “the classic adversary method” (1983, 265). Indeed, the formal adoption of majority rule developed in response to strong conflict. For example, the ancient Greeks may have used it to head off civil violence (1983, 13). As Mansbridge puts it, “a formal vote is the crucial mark of the legitimacy of conflict” (1983, 13). The English parliament adopted a formal majority vote as it became increasingly riven by deep cleavages, and by the mid-seventeenth century it “had departed enough from its traditional informal practice of unanimity to begin making decisions more than half the time by majority vote” (1983, 16). By instituting majority rule, the group acknowledges the legitimacy of conflict.⁷⁶ If the group sets out to use majority rule, and especially if it does so without the expectation of repeated unanimous votes, then it will lack a norm of developing consensus. The group is shaped by this expectation from majority rule, and the process of discussion will reflect the norm of explicit conflict. Rather than expecting to accommodate nearly everyone as much as possible, as a norm of unanimity demands, the group will assume that there will be winners and losers. Rather than banishing the very notion of faction, which the unanimity model of friendship requires, groups that rely on majority rule will tend to accept them as natural or inevitable. And instead of downplaying the role of self-interest, as the unanimity model does, the majority rule model takes it for granted that self-interest is the primary motive of the members. In groups where majority rule is common, interests are assumed to be more in opposition than reconcilable, the ground divided more than common.

The implied rule of unanimity generates a group priority on equal respect. By contrast, the implied rule of majority produces a priority on only the basic guarantee of equal individual voting power. Unanimous rule gives all the power to the unified collective; majority rule gives all the power to the more numerous faction and none to the minority.⁷⁷ Coupled with minority rights, majority rule can serve its main function of preventing the Hobbesian nightmare of a “war of all against all.”⁷⁸ But that is all it can do—it cannot encourage the listening and mutual exchange of a consensus process.

The key is that these functions of the rule are likely to shape the view of the participants of what their interaction is all about and to alter their behavior to conform to those expectations. Because “people usually adopt [a consensus process] when they expect to agree,” the person who learns that the group will use unanimous rule is likely to expect that the group will take on a cooperative orientation, then act respectfully and with the expectation of mutual respect in return.⁷⁹ In groups ruled by the imperative to include each member, norms develop that “make it difficult even to suggest that individual interests might conflict” (Mansbridge 1983, 32–33). These groups would regard bargaining, side payments, or other mechanisms that assume conflicting interests to be morally suspect (33).

And by the same token, majority rule is the best-known way of “making decisions whenever interests are both expected to conflict and do conflict on specific issues” (Mansbridge 1983, 265). In groups expecting to end their discussion with a majority vote, the members are likely to regard self-oriented acts as legitimate. Thus the presence of majority rule signals to members that the goal of discussion is to figure out what the sides are and which side will likely have its way. As Gastil puts it, “[majority rule] can lead to tense relationships among group members. Majority rule often works as a zero-sum game: one subgroup’s victory is another’s defeat. If the process becomes highly competitive, adversaries may begin to question one another’s mutuality and competence, and group discussion can turn into hostile debates” (1993, 54).

In sum, unanimous rule signals, and is likely to generate, equal status for each person, and an inclusive, cooperative norm, whereas majority rule signals the notion that the majority is entitled to get its way and is likely to produce a more competitive norm.⁸⁰

However, on their own, the gender composition of the group and its decision rule are only partial explanations. What is needed is an account of how each conditions the effects of the other. And that is what we aim to provide. We theorize, then, that decision rule and gender composition will interact to shape patterns of participation and influence within the group. This is our interaction hypothesis (see table 3.1).

The interaction hypothesis claims, first, that unanimous rule produces a group dynamic in which various types of numerical minorities—social identity as well as preference minorities—are included more than they would be otherwise. Compared to majority rule, unanimous rule benefits both genders when they are in the numerical minority of the group. But the effects of unanimity are best understood in contrast to the speech participation of each gender minority under majority rule. When women are the minority under majority rule, the gender effects we have documented will kick in, and they will speak and influence less than men in the group. But when men are the minority under majority rule, the same gender effects will restrain women from completely dominating men, so women will leverage the power of their majority only so far as equality with men in the group. Thus minority women will be included more under unanimous than majority rule, and this will shrink the gender gap in the group, rendering women more equal to men. Minority men will be also be included more under unanimous than under majority rule, and this will neutralize the potential power of women’s large numbers. Under unanimous rule the gender gap may remain unaffected as women’s numbers increase, or even grow. Unanimity will elevate both genders when they are in the minority, but women go from underrepresentation to equality, while men go from equality to overparticipation.

One possible explanation for why the elevating effect of unanimous rule produces different gender gaps is that each gender minority increases its participation under unanimous rule for a different reason. As noted earlier, in small group discussion, on average men tend toward individual agency, women toward cooperation (Miller 1985; Smith-Lovin and Brody 1989). Therefore, women in a numerical minority may interpret unanimous rule to mean that they should make at least a minimal contribution, more than in majority rule, but avoid dominating the discussion. Minority men may interpret unanimous rule as a signal that they should maximize their individual participation. That is, gender may create a bifurcated interpretation of the meaning of a consensus requirement. The reason it does so is that consensus contains within it two opposing elements. As Gastil puts it, “[consensus] radically empowers group members, often making them aware of both their autonomy and their responsibility to the group” (1993, 52). Individual autonomy exists in tension to its direct opposite, the responsibility of the individual to the collective. We argue that social identity shapes which of these two elements of unanimous rule the individual member tends to internalize. Furthermore, this effect of social identity may come into play when that social identity is made salient by the group’s gender composition. Consequently, men may view minority status as requiring maximum individual input from the minority and act on this view when they are the minority under unanimous rule. Relative to majority rule, we expect that unanimous rule elevates the participation of both gender minorities, men and women. By extension, unanimous rule helps female minorities, closing the gender gap in those groups, but it also helps male minorities, and that exacerbates the gender gap there.

Our attention to the interaction of gender composition and decision rule thus significantly qualifies the gender role hypothesis about the effect of women’s numbers, leading to our second major prediction: Women should increase their participation with greater numbers only under majority rule. That is, under majority rule, the implicit norm is that of numerical agency, and this benefits women when they are the gender majority. But under unanimous rule, greater numbers do not benefit women in the same way, because this rule aids minority men to the detriment of majority women. Succinctly stated, our interaction hypothesis is as follows. A significant interaction between gender composition and decision rule exists: the gender gap in speech and influence favoring men decreases as the number of women increases under majority rule, but remains the same or increases as the number of women increases under unanimous rule.

No interactive effect of decision rule and gender has been taken into consideration in the literatures on gender or decision rule. Yet an interaction is plausible in light of what we know about each variable in isolation. Attention to this interaction represents our theoretical contribution to these literatures.

CONCLUSION

We have laid out the general framework that guides us in the remainder of the book. Numbers matter, but their effects depend on the decision rule. For a number of reasons, women on average tend to have lower confidence and to be more affected by that lower confidence, to dislike conflict and seek cooperation, and to seek a sense of connection to others. These differences between men and women, which can be small, may become consequential when individuals assemble in groups. We identified two possible reasons why groups matter in this way. First, the fewer women are present, the more the interaction takes on a masculine character. In addition, the fewer women, the more confident participants the women encounter and the lower their sense of capacity to function as a valued member of the group. We have called this the “minority status” hypothesis. A corollary is the “enclave” hypothesis: all-female groups provide a special boost to women.

However, these effects of gender composition are qualified by the nature of the institution—specifically, the rules and procedures of interaction. Unanimous decision rule sets a norm of inclusion. That norm can be based on social bonds and mutual affinity, or alternatively, on the protection of the veto. The dynamic of affinity will help elevate women’s participation. That is because women more than men may need an interaction style that emphasizes social solidarity and cooperation, and because their lower confidence may make them especially sensitive to a dearth of, and thus especially responsive to the presence of, positive reinforcement from the group. So when women are few, they will benefit from the generally protective dynamic of consensual norms that are more likely to develop under unanimity rules. But unanimity empowers any minority group, so men, who tend to be more oriented toward individual agency, may leverage the power of the veto when they are the gender minority to maximize their participation. Consensual norms can, therefore, be a double-edged sword for women, helping them when they are the minority but not when they are the majority.

Gender is not merely a demographic characteristic of individual men or women. Gender also produces more feminine or more masculine ways of thinking, acting, and interacting. Gendered expectations in society build up lower reserves of authority for women than for men. Gender socialization leads women, on average, to have lower confidence during formal discussions, to orient more toward cooperation than conflict, and to react more sensitively to social bonds, relative to men. However, a person’s degree of masculinity or femininity is shaped by the situation. A man, or a woman, may behave in ways that are more or less masculine or feminine as the environment shifts. The expectations, norms, and habits of interaction that apply to men or to women, with their implicit codes of thought and action, nudge a person toward masculinity or femininity, but the setting can shift those expectations, norms, and habits. In that sense, settings and institutions are gendered. That is, gender is a characteristic of processes and institutions that does not necessarily refer to individual men or women at all (Winter 2008).

We identified two characteristics of the setting that help to determine whether women will experience lower authority: the number of women present and the rules and procedures that govern the discussion. The number of women is gendered in that it sets in motion signals about women’s authority and can shape the interaction to be more or less masculine or feminine. A decision rule is also gendered in that unanimous rule better matches the interaction style of women than of men, and produces more influence for women than for men. Gender is actively produced, defined, and reinforced (Scott 1986) and perhaps most powerfully when it is not noticed as such. Our contribution to this theoretical framework is to show how two gendered features of the setting or the institution—numbers and rules—jointly produce a new set of gendered dynamics.

In the following chapter we present our own study and offer methodological considerations on how to best go about testing the implications we have derived here.

 

---

¹ We make no assumptions about the sources of stable gender differences, and we are not in a position to evaluate arguments about biological versus socially constructed sources of those differences. However, the literature on gender and social behavior supports the notion that much of what we seek to understand is shaped by social processes. As Leaper and Ayres conclude, “biologically oriented researchers investigating gender-related social behaviors generally acknowledge that first, biological predispositions can be altered over time through experience, and second, existing dispositions can be mitigated or overridden by situational demands” (2007, 331). For recent work on the interaction of biology and social practices, see Eagly and Wood (2012).

² We do not wish to signal that rule is less important than composition simply by the length of the respective sections. The only reason for the relative brevity of our discussion of rule is that much less relevant scholarship has been published on it.

³ The sociological variant is known as Expectation States Theory (Ridgeway 2001; Ridgeway et al. 2009). Its core concept is “status beliefs.” These beliefs constitute a widely shared norm that identifies women (or ethnic minorities) with a particular set of skills, motivations, and abilities and grants them lower status and overall lower social worth. Joseph Berger and colleagues proposed the Expectation States Theory to explain why some group members have more influence and participate more actively than others. It argues that these inequalities are caused by and in turn reinforce the status inequalities in society more generally. We view this theory as similar to Eagly’s theory of gender roles, although there are some differences between the theories (Ridgeway and Diekema 1992).

⁴ Eagly 1987; Eagly and Wood 2012; Ridgeway 2001; Ridgeway et al. 2009. See also the literature on gender stereotypes, e.g., Glick and Fiske 1996.

⁵ People may be socialized into gender roles through self-esteem. One study found that two types of adolescents experienced a rise in self-esteem between the ages of seventeen and twenty-three: girls who viewed themselves as primarily communally oriented, and boys who viewed themselves as primarily self-oriented (Stein, Newcomb, and Bentler 1992, 465–83). These results suggest that when people conform to society’s gendered expectations, they may experience approval and their self-esteem may rise accordingly.

⁶ See, for example, Wood and Karten 1986.

⁷ As well, the gender gap in confidence is due in part to men’s overconfidence. As Fox and Lawless put it, men are “more likely to express confidence in skills they do not possess and overconfidence in skills they do possess” (2011, 61–62).

⁸ The results are also consistent with the possibility that men’s confidence drops as far as women’s, but remains higher than women’s after the negative feedback only because men begin at a higher confidence level. Regardless, the main point is that men tend to have more confidence in one sense or another.

⁹ The number of subjects is small because the study used fMRI scans, which are quite expensive. The scan results are not important to our point here.

¹⁰ An extended discussion of these issues can be found in chapter 2.

¹¹ Participants were college students, again reinforcing the notion that the gender gap in political participation is not going away on its own anytime soon, and that it is not primarily caused by women’s deficits in the workplace or earned income.

¹² Along these lines, a study of gendered speech among university professors found that women tend to use more tentative speech styles and men assertive forms. For example, men tended to say things like “I have two midterms and a final” or “I’m gonna ask you to do one midterm,” while women tended to state the same content in indirect form: “there are two papers” or “there is going to be a midterm and a final” (Tannen 1994, 175–76). The men in the study tended to make clear that they are personally issuing requirements, while the women tended to avoid doing so.

¹³ Eagly and Carli 2007; Eagly, Makhijani, and Klonsky 1992; Heilman and Okimoto 2007; Rudman 1998; Rudman and Glick 2001.

¹⁴ See also Ridgeway 2001.

¹⁵ This process is not necessarily, or even likely, reflected in explicit ideology, nor is it necessarily recognized consciously. In fact, today there exists an ideology of gender equality in the public sphere. But gender roles, and socialization into them, continue to exist.

¹⁶ To be sure, women may tend to be more concerned with others’ views of them and thus seek approval more intensely; perhaps consequently, they may try harder to avoid alienating others and maintain the connection with those in a position to judge them. But these studies also suggest that women try harder to avoid assertive communication styles because they experience negative consequences from doing so.

¹⁷ Cited in Pande and Ford 2011, 4. Original results at http://www.gallup.com/poll/24346/americans-prefer-male-boss-female-boss.aspx.

¹⁸ Even more telling of the presence of a gender norm, women behave in more self-sacrificing and self-effacing ways when observed than when in private. In one experiment, participants were told to work as long and hard as it takes to “earn” four dollars. The investigators did not define “earn,” allowing them to see if women set a higher standard for themselves than men did. They found that women indeed worked longer and harder than men, but doubly so when they were observed than when in private (Major, McFarlin, and Gagnon 1984). Women ask for lower salaries in the presence of another person than in private (Wade 2002, cited in Babcock and Laschever 2003). Women act as if they are conforming to a gendered norm of behavior. That norm teaches them to defer to others’ wishes and to avoid direct assertion of their own views.

¹⁹ As in the previous chapter, we note that these gender gaps are not going away with time. The negotiation studies find just as massive a gender gap among young people (Babcock and Laschever 2003, 67).

²⁰ They also ruled out as important factors a person’s level of risk aversion and sensitivity to negative feedback.

²¹ Belenky et al. 1986; Margolis 1992; Houston and Kramarae 1991; Noelle-Neumann 1993.

²² Dodson and Carroll 1991; Jewell and Whicker 1994; Rosenthal 1998 and 2005. For example, Rosenthal (1998) finds that women are more likely to emphasize an “integrative” style. This style is characterized by “trust, affection, a team orientation, moderation, and a commitment to process and task” (1998, 57). It highlights inclusion and relationships. She finds that men are more likely to engage in “aggregative” leadership that features “dominance, competitiveness, ambition, a drive to control, and opportunism” (1998, 57).

²³ McKinsey 2008 cited in Pande and Ford 2011, 21.

²⁴ See, for example, Walker 1984 and Jaffee and Hyde 2000 cited in Huddy, Cassese, and Lizotte 2008, 39.

²⁵ Aries 1998; Babcock and Laschever 2003; Babcock et al. 2003; Cross and Madson 1997; Eagly 1987; Eagly and Johnson 1990; Knight and Dubro 1984; Josephs, Markus, and Tafarodi 1992; Leventhal and Lane 1970; Scott et al. 2001; Schwartz and Rubel 2005; Sidanius and Pratto 1999. Studies find this across many countries (e.g., Schwartz and Rubel 2005) but not in all (Prince-Gibson and Schwarz 1998).

²⁶ Even when men took the liberal position, they tended to introduce bills that focused on rules regulating infringement on individual rights, such as those governing new evidence in criminal cases. While some women sponsored bills that toughened policy, only women introduced bills that focused on prevention.

²⁷ Some studies discuss what goes on in group discussions but have not systematically studied the effects of different characteristics of groups and have not focused on gender (e.g., Walsh 2007; Gastil 1993, 2010).

²⁸ See, for example, Aries 1996, 1998; Deaux and Major 1987; Eagly 1987; LaFrance, Hecht, and Paluck 2003; Leaper, Anderson, and Sanders 1998; Leaper and Smith 2004; Ridgeway and Smith-Lovin 1999; Eagly, Wood, and Diekman 2000; Leaper 2000.

²⁹ Here we draw on the general assumptions of Expectation States Theory (Ridgeway 2001; Ridgeway et al. 2009) and on the interactionist theory of gender, which argues that gender roles become more or less salient with the situation (Eagly and Carli 2007).

³⁰ Aries 1976, 1996; Hannagan and Larimer 2010; Johnson and Schulman 1989; Smith-Lovin and Brody 1989; Mendelberg and Karpowitz 2007.

³¹ Boyd, Epstein, and Martin 2010; Farhang and Wawro 2004, 2010; Grünenfelder and Bächtiger 2007; Massie, Johnson, and Gubala 2002; Peresie 2005. While few strong or consistent gender effects are evident in individual-level analyses of the decisions of women serving on the bench (Ashenfelter, Eisenberg, and Schwab 1995; Segal 2000; Songer, Davis, and Haire 1994; Walker and Barrow 1985), much larger effects emerge when examining the gender composition of judicial panels. These effects of gender composition appear to hold even after controlling on ideology and other factors. In a sample of cases over a nearly twenty-year period, Massie, Johnson, and Gubala (2002) find, for example, that judicial panels with at least one woman are more likely to take pro-plaintiff positions in criminal procedure and civil rights cases. Similar gender-composition effects show up in Farhang and Wawro’s (2004) study of employment discrimination cases, with panels that include women again proving more likely to take pro-plaintiff positions. These effects extend to the individual judge’s opinions as well as the overall decision of the panel. The most powerful effects are reported by Peresie (2005), who finds that in sex discrimination and sexual harassment cases, the more women on the panel, the more likely the panel is to find in favor of the plaintiff, even after controlling for ideology and a variety of other factors. More sophisticated and rigorous studies echo these findings (Boyd, Epstein, and Martin 2010; Farhang and Wawro 2010). These studies and others support Sapiro’s general conclusion: Differences between the average man and woman may be small and inconsistent, especially in comparison to large variances within gender, but these differences can become large and consequential when amplified by group-level forces (Sapiro 2003). It is not the case, though, that we already know that gender composition increases ordinary women’s participation and influence, as we explain in the appendix to chapter 4. In addition, there is little work on the mechanism explaining gender composition effects or the process by which judges interact.

³² Aries 1998; Bowers, Steiner, and Sandys 2001; Croson and Gneezy 2009; Eagly 1987; Giles et al. 1987; Hastie, Penrod, and Pennington 1983; Huckfeldt and Sprague 1995; Ridgeway 1982; Strodtbeck, James, and Hawkins 1957.

³³ Another study, by Barbara Ritter and Janice Yoder, also finds that even women with a dominant personality are unlikely to assume leadership roles in dyads when paired with a man. The researchers assigned unacquainted students to task pairs, such that one member was selected based on having a higher dominant-personality score than the other member. They found that in same-sex pairs, the dominant member emerged as the leader of the pair. But in mixed-gender pairs, dominance had no effect, and the man was more likely than the woman to emerge as leader regardless of their dominance level. So even when women are inclined toward leadership by having a dominant personality, they will be less likely to act accordingly than men are (Ritter and Yoder 2004).

³⁴ Women may also wish to avoid outperforming men out of a fear that they will be rejected as romantic partners. Two female MBA students at Harvard on CBS’s 60 Minutes stated in 2002 that they no longer tell men that they go to Harvard so as not to squash potential romantic relationships (cited in Babcock and Laschever 2003, 103).

³⁵ Djupe, McClurg, and Sokhey 2010; Huckfeldt and Sprague 1995. In addition, in studies of actual car sales interactions using trained confederate consumers, women are quoted higher prices than men in anticipation of women’s weaker negotiation behavior, and in lab studies of ultimatum games, women receive lower offers from their partners than men do and are expected to make higher offers than men (Babcock and Laschever 2003).

³⁶ Pugh and Wahrman 1983; Ridgeway 1982; Wagner et al. 1993; see the review by Foschi and Freeman 1991.

³⁷ Similar implications derive from research on social dominance orientation (Pratto, Sidanius, and Levin 2006) and on sexism (Glick and Fiske 1996).

³⁸ Aries 1998; Carbonell 1984; Davis and Gilbert 1989; Nyquist and Spence 1986.

³⁹ For statements of this notion that rest on the claim that women have less power and status in society, see Kanter 1977a; Lakoff 1975, 1990; O’Barr 1982; O’Barr and Atkins 1980.

⁴⁰ One may wonder about the threat hypothesis, which would predict that increasing numbers of women provoke a sense of threat among men and a backlash against women. However, a recent review of the literature in workplace settings concludes that there is no strong evidence for a threat thesis when it comes to women’s level of participation or evaluations of women; it is low rather than high numbers that hinder women. The authors conclude: “many studies are preliminary, and there is rarely a set of consistent findings based on well-specified models from which to draw final conclusions” (Reskin, McBrier, and Kmec 1999, 55.)

⁴¹ According to Schmitt and Hill’s study, the greater white men’s share of the group, the more positive were evaluators’ assessments of other individual white men. There were no effects on evaluations of white women, but the evaluations of black women in the group were negatively and significantly correlated with the number of white men in the group.

⁴² Aries 1976; Dindia and Allen 1992; Ellis 1982; McCarrick, Manderscheid, and Silbergeld 1981; Miller 1985; Smith-Lovin and Brody 1989; Mendelberg and Karpowitz 2007.

⁴³ We discuss the case of masculine female leaders in the conclusion chapter.

⁴⁴ An overview of the literature on work groups concludes that “women in the minority in their work groups felt more isolated than either women who were in the majority, or men regardless of their share of group members” (Aries 1976, 15; Carlock and Martin 1977; Reskin, McBrier, and Kmec 1999, 346).

⁴⁵ For this general argument see Burns, Schlozman, and Verba 2001.

⁴⁶ Hannagan and Larimer argue from an evolutionary perspective that group composition matters because women are more likely to seek consensus in a group and are more likely to both (indirectly, nonverbally) signal their wishes and to read those of others. Thus groups with more women are more likely to effectively converge on the median. See Hannagan and Larimer 2010.

⁴⁷ Habermas believes that such spaces are still necessary and laments what he sees as their erosion by the mass media and large, powerful organizations. These entities, he believes, attempt to manufacture consensus through advertising and public relations. In addition, by preempting face-to-face meetings, they rob people of the opportunity to develop their ability to discuss notions that check domination.

⁴⁸ Theda Skocpol writes that for many decades in the nineteenth century, educated, higher status women were “mainstays of voluntary membership federations” (Skocpol 1999, 482). She also argues that the associations of that time were much more mixed along class lines, though much more segregated along lines of race.

⁴⁹ There is also a literature on single-sex education. It offers mixed results and suffers from a short-age of rigorous studies. For an example of positive findings regarding women’s colleges, see Smith 1990. Some reports find that women’s colleges tend to engage in a concerted program of leadership education for its students, presumably more so than co-ed colleges do, though that is not tested (Whitt 1994). A recent report by the US Department of Education concluded that girls in all-girl schools tend to have a higher level of academic achievement and career aspirations. However, many of these studies are confounded by comparing Catholic with public schools. http://eric.ed.gov/PDFS/ED486476.pdf.

⁵⁰ For extensive documentation of the role of local meetings in women’s groups around the United States, see Skocpol 1992.

⁵¹ Minkoff, cited in Skocpol 1999, 470.

⁵² Before asking gender composition, they also filtered out people who were not at least minimally active in the organization by asking if the person had donated at least $25 or given some time to the organization (Burns, Schlozman, and Verba 2001, 229, note 7).

⁵³ These percentages are 2 points different in table 9.4 (page 228), perhaps because of the treatment of missing responses—19, 15, and 62, in order.

⁵⁴ For men: 24% of all-male groups are service/fraternal or veterans’, 31% are unions/business/professional, 16% are hobby/sports clubs. All-female groups: 16% service/fraternal, 13% religious, 10% business/professional, 9% educational, including PTAs (Burns, Schlozman, and Verba 2001, 228, note 6).

⁵⁵ Percentages are of full-time employees.

⁵⁶ Eighty-seven percent of businesses had fewer than twenty employees (Reskin, McBrier, and Kmec 1999, 346, citing US Census Bureau).

⁵⁷ See also the sociology literature: Smith-Lovin and McPherson 1993, page 238; Mcpherson and Smith-Lovin 1986; McPherson and Smith-Lovin 1982; McPherson and Smith-Lovin 1987; Ridgeway and Smith-Lovin 1999.

⁵⁸ The numbers in this section are drawn from table 9.5, page 230, in Burns, Schlozman, and Verba 2001.

⁵⁹ Here, we are comparing women in all-female groups to men in all-male groups.

⁶⁰ Whether men, too, benefit from their own enclaves is a possibility we can examine in our data. The evidence from existing studies is inconclusive. For example, in contrast to Burns, Schlozman, and Verba’s positive finding, according to Bird and Wharton (1996), men in predominantly male and mixed-sex work groups were more satisfied than men in all-male groups. All-male groups exhibit the highest levels of dominance behavior (Aries 1976). However, Smith-Lovin and Brody found that men offer more positive interruptions to other men in all-male than in mixed-gender groups (1989). These findings are not directly contradictory, of course.

⁶¹ The most discussed ascriptive variables in studies of inequality and political participation are age and race/ethnicity. The voting turnout gap between youngest and oldest is about 20 points, and the gap between the most- and least-participating racial groups is 15 points (reported by Michael Macdonald, based on the 2008 Current Population Survey. The percentages for the age and the race gap, respectively, are 22 and 20 in 2004. http://elections.gmu.edu/CPS_2008.html). The biggest effect in the whole literature is for education, which moves the probability of voting by approximately 45 points. The gender gap in voting turnout is much more modest; men and women differ far less than the young and old or white and Latino Americans. Yet as powerful as age and race are in American political participation, far more powerful is the effect of gender composition on authoritative participation in civic groups. And that effect dwarfs the effect of individual gender. However, we do not wish to put much weight on this contrast, since comparing the effect of composition on organizational participation against the effect of other demographics on voter turnout is comparing two different predictors on two different outcomes. The ideal comparison would be on the same participatory variables, but we are not aware of findings reporting those. In any case, our point is that no matter what type of participation is examined and what type of predictor, gender composition effects are large by comparison.

⁶² When we compare female enclaves to male enclaves, the pattern is also consistent with the notion that enclaves strengthen women’s authority. There are two experiences that produce more than a 3-point gap favoring women when we compare all-female and all-male groups (we set a gap of more than 3 percentage points as our threshold here as a proxy for statistical significance): Opinion was asked, and feel some control over policy. Women’s enclaves help women (above the help that male enclaves give to men) in these two respects. One of them implicates the exchange of opinions—women’s opinion is solicited more in all-female groups than men’s is in all-male groups. The other is a general sense of empowerment.

⁶³ Aries 1976.

⁶⁴ For observational or quasi-experimental studies, see Carlock and Martin 1977; Reskin, McBrier, and Kmec 1999, 346.

⁶⁵ Buck et al. 1972; Hall 1984; Wagner, Buck, and Winterbotham 1993; for a nonexperimental study, see Falk 1997.

⁶⁶ We do not mean to overstate the case for ease. Reaching agreement on a decision rule can lead to conflict and confusion, and there may be slippage between the formal rule and its implementation (Gastil 2010, 101).

⁶⁷ Gastil 1993; Mansbridge 1983.

⁶⁸ Gastil 1993.

⁶⁹ Although the results are muddied by the fact that these juries engaged more frequently in various forms of speech.

⁷⁰ Frolich and Oppenheimer 1992; Kameda 1991; Kaplan and Miller 1987; Nemeth 1977.

⁷¹ Some studies qualify these positive effects, finding that majority rule better neutralizes the power of a dominant member (Falk 1981), perhaps only when that member is selfish (Ten Velden, Beersma, and De Dreu 2007).

⁷² Falk 1982; Gero 1985.

⁷³ See, for example, Gastil 1993, 50–53.

⁷⁴ On the former claim, see Miller 1989; Mendelberg 2002; Devine et al. 2001.

⁷⁵ This discussion draws heavily on the ideas in Mansbridge (1983).

⁷⁶ As Mansbridge writes, “By accepting some conflict as legitimate and by instituting the formal procedures of one citizen/one vote and majority rule, Athens became the first society to move away from unitary democracy” (1983, 15).

⁷⁷ On the notion of faction, consider Aristotle’s view that in the consensus-oriented group, members “will not tolerate faction at any cost” (quoted in Mansbridge 1983, 14).

⁷⁸ Mansbridge and Gastil tend to distinguish between a consensus process versus a formal unanimity voting rule (Gastil 1993, 50–53; Mansbridge 1983). However, in our view the distinctions between the formal rule and the actual process are not as important as the commonalities, and we view the two as more overlapping than distinct. We suggest that the unanimous voting rule tends to signal the need to engage in a consensus process. We grant Mansbridge’s point that a formal unanimity rule is likely to signal a less consensually oriented process than the absence of any formal rule, and that a formal rule may be used at both ends of the conflict spectrum (1983, 33). The most prominent example Mansbridge notes is the UN Security Council. This body uses unanimous rule in the sense of giving a veto power to each member, but that institution does not impose a sense that it is illegitimate for a member country to act exclusively in its self-interest. Similarly, economists tend to view unanimous rule as a rule like any other, one that generates individualistic strategic behavior rather than transformative cooperative norms (Austen-Smith and Feddersen 2006). However, these claims do not materially affect our argument here, which is that adopting a unanimous rule is likely to move the group further along toward a norm of cooperation and inclusion. That claim assumes a relative movement from what may be a very low level of cooperation. Such an effect may apply to a highly contentious group whose members are constrained from developing meaningful friendship as well as to the closely bonded group whose members assume the existence of shared basic interests (Gastil 1993; Mansbridge 1983). Some of Mansbridge’s writings are consistent with our argument, as when she argues that unanimity often serves the function of protecting interests from tyranny and of building bonds of amity and commonality (1983, 263).

⁷⁹ The quote is from Mansbridge (1983, 32), but the idea that rules create norms, while drawing on her discussion, is primarily our conjecture.

⁸⁰ To be sure, the mere existence of a formal vote is not as important as whether the group develops a norm of orienting to consensus. As Mansbridge puts it, “Although the [Athenian] assembly used majority rule, it may well have made most of its decisions by consensus” (1983, 14). That is, groups may use the option of taking a majority vote yet still rely on a pro forma vote for nearly all decisions, with the expectation of unanimity in general. In that case, the actual decision-making procedure is unanimity rather than majority rule. In such a group, for any given discussion the likely expectation is of a unanimous vote, if a formal vote is to be bothered with at all. In such groups, the de facto rule becomes unanimous rule.