ABSTRACT

As predicted by resource-based models of political participation, women's increasing education and labor force involvement have led to rates of political activity equal to or exceeding rates for men. But NES election study data reveal that a sizeable gender gap has persisted from 1972 to 1994 in efforts to influence others' votes. An alternative contextual model is postulated to account for increased levels of political persuasion efforts by women when female candidates are on the ballot for major elective office. Logistic regression is used to predict whether or not a respondent will attempt to influence others' votes, and whether a contextual variable (presence of female candidates in the respondent's state or Congressional district, 1990-1994) improves the prediction. In 1992, women candidates were associated with higher levels of political involvement, internal political efficacy, and media use by both sexes, but the effects of candidate gender were stronger for women. In the Congressional elections of 1990 and 1994, candidate gender had no such impact; partisanship and political involvement remained the strongest predictors of political proselytizing.
 

--------------------------------------------------------------------------------

Who is she, what does she want, and what sort of action shall she undertake? Is she a human being who can give a spoken account of herself? If so, what language does she speak?
- Jean Bethke Elshtain (1982: 614), on feminist discourse
 

Do women speak "In a different voice" in politics? Men's and women's political attitudes indeed differ on issues such as the use of force (war, crime, gun control, capital punishment), concern for the environment, and support for social programs as opposed to defense spending (Mueller 1988; Cook and Wilcox 1991; Kathlene 1995; Hurwitz and Smithy 1996). Although women vary by race, class, religion, and other characteristics, as a group they are more likely than men to be strong partisans, and tend to be more Democratic and liberal in their voting patterns. Schlozman et al. (1995) found that educational issues and abortion were especially salient on the policy agendas of female activists.

Despite these differences, women's "different voice" (as Gilligan (1982) termed it) may well be muted because women are less likely to engage in political discourse. Women discuss politics less often with friends and family (Huckfeldt and Sprague 1995, Ch. 10), and have been found to be significantly less likely than men to report efforts to try to convince others how to vote (Welch 1977; Rapaport 1982). At least before the 1980's, this pattern held up regardless of race, level of education or political involvement; in fact, gender differences in political proselytizing were largest among the college educated in 1972 (Hansen et al. 1976.) Women of course have long been active in other ways not always tapped by conventional research on political behavior: education, union activity, consciousness-raising, grass-roots advocacy, volunteer work (Bookman and Morgen 1988; Kathlene 1989; Fowlkes 1992), and especially religious groups (Schlozman et al. 1994). But if election outcomes do matter for women as they do for men, women's informal role in campaigns merits our attention.

This paper will explore gender differences in political proselytizing using National Election Study data. As we shall see, gender differences in this type of political conversation have persisted from 1972 to 1994, despite far-reaching changes in women's employment and education, and diminution of gender differences in other forms of political participation. However, in 1990, 1992, and 1994, the presence of visible female candidates for public office affected the political interest and involvement of both men and women - but especially women. Logistic regressions show that in 1992, men were somewhat less likely to engage in political proselytizing if women were on the ballot, while women were significantly more likely, despite controls for partisanship, age, party, and issue positions. These results suggest that political context affects the impact of gendered norms on political discourse.
 

1. The Importance of Political Proselytizing
Why is political proselytizing important? The Greek philosophers held that only by participating in political discourse in the public forum could a person (that is, a free adult male) become fully human. Modern theorists such as Locke, de Tocqueville, Mill, and Dewey have also stressed the contribution of free and open political discussion to democracy (see historical summary by Kinder and Herzog 1993). Teachers often learn through their attempts to communicate their knowledge and ideas to others. Earlier voting studies emphasized the two-step flow of communication (Berelson et al. 1954; Robinson 1976): events (and media coverage of those events) have a greater effect on people when they are reinforced by the interest and opinions of our friends, neighbors, and coworkers. For example, as Katz and Feldman (1962) found, the Kennedy-Nixon debates had the greatest impact, not on the partisan faithful who actually watched them, but on the people who learned about them from their friends' reactions. Word-of-mouth sells more movie tickets than do favorable reviews or big advertising budgets (Wyatt 1994), and one would expect a similar relationship for political ideas and candidate preferences.

Yet many Americans are not comfortable discussing politics, and are hesitant to attempt to influence others' views. NES post-election survey results show that, on average, only about a third of the population reports any effort to influence others' votes (even fewer in midterm elections.) In part, this stems from low levels of interest and engagement in electoral politics; people with minimal political involvement are unlikely to initiate political discussion. Aversion to conflict is another possible explanation for people choosing to "clam" rather than talk. As MacKuen (1990) argues, expectation that one's interlocutor is a kindred spirit will increase the likelihood of political self-expression (see also Noelle-Neumann 1993). But any attempt to influence another's vote risks offending someone, or politicizing a neighborly or work relationship, or causing a family argument -- precisely the reasons polite Victorians eschewed discussion of politics along with sex and religion. Knowledgeable and self-confident citizens are more willing to take that risk; political proselytizing (and political discussion more generally) are more frequent among the well educated and those with high levels of self-esteem (Bennett et al. 1994).

American social norms impede political discussion by many men as well, but may be particularly problematic for women. Although women's voting rates now exceed men's, women have usually been found to have less political interest and involvement than men (Bennett and Bennett 1989). Political conflict may be particularly unpleasant for women socialized by traditional feminine gender-role norms to be sensitive to others' feelings and to maintain relationships (Belenky et al. 1986). Noelle-Neumann`s analysis of the social context of public opinion found that "men are more disposed to join in talk about controversial topics than are women" (1989: 24), although both men and women were more comfortable speaking up if they sensed agreement with their fellow conversationalists. Women are also underrepresented on talk radio, an increasingly important factor in elections and public opinion in the U. S. A report by the Times Mirror Center for the People and the Press (1993) found that men outnumbered women two to one among callers to radio talk shows. "Conservative, vocal, angry men" were far more prevalent on the airwaves than liberals or feminists.

Several recent feminist analyses have suggested that the content and style of political discourse is alienating to many women (Elshtain 1982; Tannen 1994). Jane Margolis (1989) found that, at least for a sample of Harvard undergraduates, "gov talk," with its depersonalized abstractions and rationalism, made many women students uncomfortable, less likely to speak up in class, and unwilling to challenge others' views in public (1). Rapaport (1982) found that while adolescent males and females were equally likely to express opinions on non-political issues, females were more reticent about political ideas and values. Women were less likely than men to feel that they understood political issues, even when their objective level of political information was high. Rapaport's analysis suggested that these socialization patterns persist into adult life. Kathlene's (1994) discourse analysis of Colorado legislators found that successful women legislators, while hardly reticent, used somewhat different symbols, images, and linguistic conventions than their male counterparts.

If women, for whatever reasons, do not take part in the flow of interpersonal political communication, their opportunity to exert influence over candidates and issues will be reduced. Their socialization and lack of resources (economic, social prestige) may also render them more vulnerable than men to influence by others. As Mueller (1991: 34) found, "occasional woman voters" in 1988 were preoccupied with immediate family concerns, seldom followed politics regularly, and were most vulnerable to the Bush campaign's emphasis on emotional issues such as the flag, patriotism, and Willie Horton. In Sapiro's (1983) insightful terms, many American women behave more as subjects than citizens, and are motivated more by a sense of civic duty than by commitment to issues or candidates. Although many women (in recent years, even more than men) do vote, fewer engage more actively in the political process. Political activity undertaken in the isolation of the voting booth can have little influence on others.

2. Gender and political proselytizing, 1972-1994
If political participation is indeed a function of individuals' resources, we would expect the gender gap in political proselytizing to have diminished over time as women's access to education and prestigious occupations has increased. But that is not the case. Table 1 shows responses by sex, 1972-1994, to the National Election Study's repeated question "During the campaign, did you talk to any people and try to show them why they should vote for or against one of the parties or candidates?" (2) (data from Miller 1994; Rosenstone et al. 1995.) Presidential and midterm elections are grouped separately in Table 1, because it is readily apparent that more political proselytizing occurs in Presidential years among both men and women, when interest in the campaign is higher. Overall, even in Presidential years, only about a third of the electorate reports any attempt to influence another's vote, and in off-years this drops to 20 percent. But in every single election year since 1972, men have been more likely than women to report attempts to influence others' votes, and this gender gap was even larger in 1992 than it was in 1972. The gender difference averages seven percent in mid-term elections, but fluctuates between four and twelve percent in Presidential years.

[ Table 1 about here ]

The high points in this series occurred in 1976 and 1992 among Presidential elections, and the nadir was 1988 (a year in which the absence of issues became itself an issue; Elshtain 1989.) Before 1994 there was some evidence for a decline over time in the already low rate of political conversation in Congressional election years; in 1990 only 13 percent of women and 22 percent of men reported any attempt to influence others' votes (3). In the precedent-shattering 1994 midterm election, men's rate of proselytizing was considerably higher than usual, but women's only marginally so.

These patterns are quite different from those observed for other forms of political participation by men and women. Indeed, male-female differences in voting and campaign activity had largely disappeared by the 1980's (Beckwith 1986.) Rosenstone and Hansen's analysis of NES data, 1956-1988, found that gender was a significant predictor of political proselytizing but not of other electoral activities (1993: 131.) Bennett and Bennett (1989) suggest that this is because political proselytizing is closely linked to interest in politics and in campaigns, and women have consistently been less politically interested than men.

Verba et al.'s (1995) comprehensive analysis of political participation found moderate gender differences persisting in campaign contributions, contacting government officials, and affiliation with political organizations (1989 data), but they did not consider attempts at political persuasion. Their "resource model" of political participation would suggest that as individuals acquire more civic skills (through education, work experience, or organizational involvement) their likelihood of political engagement should increase. But gender differences in political proselytizing do NOT appear to track ongoing secular changes in women's educational or occupational status. More women have attended college (42 percent of the NES sample in 1992 compared with 26 percent in 1972) and many more women have entered the labor force. Nevertheless, a continuing "gender gap" in political proselytizing is evident even among working women, college- educated women, and those with professional or managerial jobs. In 1992, the proportion of college-educated women who attempted to influence others' votes (42 percent) was exactly equal to the proportion of high-school-educated male political proselytizers. In 1992, 24 percent of all political influencers were men with at least some college, although such men constitute only 16 percent of the total population, and this pattern has changed little since 1972.

Why have economic and demographic trends NOT overcome women's reluctance to try to convince others how to vote? First of all, access to educational and occupational resources still differs by gender. Women may be receiving more years of education, but as many feminist scholars have argued, the content of that education, and the social setting in which it occurs, continue to devalue women's experiences (Bailey et al. 1992; Margolis 1992). Although many more women are in the labor force, occupational segregation on the basis of sex has changed little since the 1970's (Reskin 1993), and working women still do most of the housework, leaving less time for political involvement. Women consistently earn less than men for full-time, year-round work, and relatively few women have moved into the highest levels of managerial or professional positions (Goldin 1990). Thus having a college degree, or holding a job, may not necessarily afford women the opportunity to acquire the social skills or the self-confidence that encourage men to proselytize. Schlozman et al. (1994) find that women's lack of access to economic resources accounts for their lower rate of campaign contributions. Political proselytizing does not require money, but the social power and prestige that money conveys may make it easier for men to proselytize.

Second, socialization patterns that reinforce traditional gender norms concerning politics, discourse, and social conflict (Kramarae 1982) can reinforce women's lack of access to social status or economic power. As Kanter (1977) and others have found, women tend to be the peace-keepers and consensus-builders in social and business situations. Margolis (1992) and Belenky et al. (1986) suggest that women are especially likely to remain silent if they anticipate opposition (see also Noelle-Neumann 1993; Houston and Kramarae 1991.) Sapiro (1983) described women as better "subjects" than citizens because of their preference for the more passive (and less conflictual) forms of political engagement.

A third approach, however, would place less emphasis on the characteristics of the electorate and more emphasis on the cues provided by the political context. In contrast to research on atomized individuals acting in isolation, contextual models stress the importance of social setting and social interaction on people`s attitudes and behavior (Books and Prysby 1988; Jelen 1992; Huckfeldt and Sprague 1991, 1995). From this perspective, women would be less likely to attempt to influence others' votes in the absence of issues or candidates that specifically engage their interests. As Rosenstone and Hansen (1993) show, the mobilizing strategies of politicians and parties interact with the resources of individuals to explain the dynamics of political participation. They further suggest that the predominance of white male politicians will discourage political communication by women and blacks (1993: 78). Bobo and Gilliam (1990) demonstrate that the presence of African-American mayors in large cities significantly affected the political involvement of African-American citizens. They suggest Black elected officials provide a "contextual cue of likely political responsiveness that encourages blacks to feel that participation has intrinsic value" (1990: 387; see also Tate 1991). Hero (1992) makes a similar argument concerning Latinos.

Could female candidates have a similar impact on women? Consider 1984, the only year in which the gender difference in political proselytizing was NOT statistically significant at the .01 level, and was the smallest in absolute terms. In that year, women reported attempts to convince others how to vote at a rate a little more than average, but men's rates were lower than usual (although still higher than women's.) The survey evidence makes clear that Geraldine Ferraro's controversial vice-Presidential candidacy did not help the Democrats, and that women were not more likely to vote for her; as in most elections, the vice-presidential candidate had only a marginal impact on the overall electoral outcome (Keeter 1985; Miller 1988). But as Steinem (1986: 394) suggests, on the basis of CBS/New York Times exit polls, Ferraro's candidacy galvanized many women into political activism that moved beyond simply casting a ballot:
 

"Ferraro served to bring out more voters, enthusiasm, volunteers, and money among Mondale's natural constituency. There is evidence that her presence was a net plus. 23 percent of all adult women said that Ferraro's candidacy had made them more interested in politics. That translates to 21 million women."
The 1984 shrinkage in the "gender gap" in political persuasion suggests that the presence of a woman on the ballot may indeed affect political discourse by both men and women. But an adequate test of that hypotheses requires more than exit-poll evidence from one vice-Presidential candidacy. Until recently, however, the number of women running for Governor or Congress has been small. And in some years when women did run for Senate (Nancy Landon Kassebaum in 1978, Barbara Mikulski in 1990), the NES sample did not include respondents from their states. But the number of female candidates increased sharply beginning in 1990. The 1992 "Year of the Woman" produced record gains for (mostly Democratic) women in Congress, and the number of women running and winning remained high (although considerably more Republican) in 1994 (CAWP, 1994). We can therefore make use of NES election studies from the 1990's to test for the impact of women on the ballot (4).

3. Candidate gender and political proselytizing
Although women's representation at state and local levels has increased steadily since 1970 (Carroll et al. 1991), there have been few opportunities at the national level to test for candidate gender as a contextual variable affecting political involvement or participation. True, a handful of women have held House or Senate seats, but Hansen (1996) found no impact of these women on rates of political proselytizing prior to the 1990's. Given their small numbers, few voters could expect them to have much impact on women's interests or issues. Until quite recently, most women elected officials found it prudent to downplay any identification with "women's issues" or feminism (a pattern that persists in more conservative states and among Republican women) (Thomas 1994). 1990 produced a sharp increase in the number of women running for Congress and for state executive office, but their rate of success remained low, largely because their opponents were usually incumbents (Darcy, Welch, and Clark 1994).

1992 was widely touted in the media as the "year of the women," and indeed many more women ran that year for Congress and for state office (Wilcox 1994). The televised Clarence Thomas/Anita Hill hearings before the all-male Senate Judiciary Committee dramatized the low rate of female representation in the Senate. Growing anti-incumbency sentiment and the 1990 redistricting encouraged more women to run and improved their chances of winning. Domestic issues (particularly abortion: Abramovitz 1995) were salient. As Mandel (1993) argues, the pool of qualified, experienced, well-funded women candidates had grown steadily throughout the 1980's. We might therefore expect that by 1992, even more than in 1990, the presence of a woman on the ballot as a major-party candidate for statewide or national office might indeed attract the attention of voters, especially women voters, and encourage them to attempt to influence others' votes.

To test for this, Table 2 compares differences in political proselytizing among men and women in states with and without a female major-party candidate for the Senate or Governor, and House districts with and without female candidates in 1990, 1992, and 1994. The combined impact of female candidates for the three offices is also computed. The unique 1992 race in California--two Senate races in the same year, both with women candidates --is shown separately. Because incumbency is so important in House races, that factor is included to test whether political persuasion is more likely if the female candidate is an incumbent (and presumably better known); this does appear to be the case (5).

[ Table 2 about here ]

As Table 2 shows, having female candidates indeed makes a difference; most of the relationships are positive and stronger for women than for men. Most dramatically, in 1992 the historic difference in male and female rates of political persuasion persisted in states with all male Senatorial candidates, but was reduced when women were on the ballot. In California, the level of political proselytizing among both men and women was well above that of states with only one woman running for Senate, and women were more likely than men to attempt to sway others' votes (6). But the same pattern does not hold for 1990. Although there were more female Senatorial candidates (seven) than ever before in that year, the number was still low, it was a non-Presidential year, and the media hype resulting in part from the Anita Hill/Clarence Thomas hearings had not yet occurred. In fact, both sexes reported FEWER attempts to influence others' votes in states with female candidates running for the Senate. The relationships for Senate races are again positive for both sexes in 1994, although only modestly so.

The Congressional pattern is very different. In 1990, 1992, and 1994, women are marginally more likely to attempt to influence others' votes when a woman is on a major-party ballot in their Congressional district. However, the same pattern holds for men in both 1990 and 1992. But regardless of the gender of the Congressional candidates, women still discuss politics at a lesser rate than men do. With governors, the small number of states in 1992 limits confidence in the results, but in both 1990 and 1994 the presence of a woman gubernatorial candidate had a significant impact on women's attempts to influence others' votes.

A combined effect of female candidates for these three races is also apparent, although much stronger (especially for women) in 1992. In that year, the gender gap in attempts at political persuasion all but disappears if two or more women are running. Overall, men are slightly more likely to discuss politics if women are on the ballot, but women are significantly more likely to do so.

These results hint strongly at the influence of contextual factors on rates of political persuasion. But another contextual effect should first be considered: do the populations of states where women are nominated for major office differ along some dimensions (levels of education, racial composition, political culture, age of population) that might produce higher levels of proselytizing whether or not women were on the ballot? A comparison of median education levels for the NES samples found no significant difference between states that did or did not nominate women. California has a large sample interviewed (N=243 in 1992), but this very diverse state parallels the national sample in terms of age structure, ethnicity, and level of education. Thus demographic factors cannot account for the state's high levels of political discussion in 1992.

A full test of state-level differences is not possible because the NES sample included only 33 states in the 1992 study, and its national sample is not designed to represent particular states. Many smaller or rural states (which tend to elect fewer women to state legislatures) were not in the NES samples in any of these three years (7). But when I compared rates of political discussion in 1988 among the ten states that nominated women for the Senate in 1992, in neither year did such states show higher levels of political discussion by men than the national average. Rates for women were very slightly higher than average in 1988, but still much lower than men's, and much below rates for those same states in 1992.

4. Logit analysis of impact of female candidates
The presence of women on the ballot thus emerges as a plausible hypothesis for gender differences in rates of political proselytizing. But we also need to weigh other factors suggested by previous research as reasons why some people attempt to sway others' votes. The first is that of individual resources. Persons who are better educated, better informed, and in possession of good verbal skills are more likely to attempt to persuade others, just as they are more likely to engage in other forms of political activity. Thus we would expect whites to be more likely to discuss politics than blacks, and Republicans more than Democrats, in each instance because of difference in access to resources. The second factor is that of psychological involvement in politics or in a particular campaign. Strong partisans, ideologues, and those who play close attention to politics are more likely to try to influence others' views than are independents or the politically apathetic. Third, as Rosenstone and Hansen (1993: 138) found, younger people are more likely than older ones to engage in this type of political activity.

These explanatory factors will be analyzed using logistic regression, the appropriate methodology when one wishes to regress a dichotomous dependent variable (in this case, whether or not the respondent attempted to influence others' votes) on a series of independent variables (Aldrich and Nelson 1984). Unlike ordinary least-squares regression, logistic regression assumes that each of the explanatory (independent) variables depends on the values of the others, and thus allows us to predict the odds of a certain category of respondent engaging in political persuasion (Norusis 1990). Coefficients for sex, race, partisanship, party identification, political interest, age, and education will be compared to those for the presence of women candidates on the ballot in a respondent's state or Congressional district (see Appendix for coding of these variables.) Since, as I have shown, the impact of women candidates is larger for women, Table 3 includes an interaction term to estimate this effect in 1990, 1992, and 1994. And to further rule out state social or cultural characteristics as predictors of higher levels of political discussion, we include in 1988 a dummy variable to indicate whether or not a particular state or district nominated a woman in 1992, as well as an interaction term for this variable and gender.

[ Table 3 about here ]

The results show that the presence of women candidates does have an impact when other factors are controlled- but only in 1992. There is no significant impact of either the female-candidate variable or the interaction term in 1988, 1990, or 1994. In all four years, the presence of women on the ballot meant, overall, less likelihood of attempts to influence others' voters. But for women in 1992, as the coefficient for the interaction term indicates, the odds of political influence attempts were indeed higher for women if women were running, even when controlling for other factors.

Although most of the women running were Democrats, party identity was not a good predictor of political proselytizing in 1992 (or in 1994): in both of those years Independents did more of the talking. In 1992 as in 1988 younger people (the omitted reference category) were more likely to proselytize; the coefficients for middle-aged and older persons are both negative in Presidential years as higher-salience campaigns draw more people into the political arena (8). Males, whites, and the better educated are consistently more likely to try to influence others' votes; the increasing negative coefficients indicate that, since 1988, whites have become even more likely to proselytize than blacks. In all four election years, a respondent's level of interest in politics and in the campaign was the strongest predictor of political proselytizing, but in 1992 gender differences and partisanship counted as well. 1992 also stands out because this model makes more errors in classifying cases in that year, suggesting that one must look beyond the usual factors to explain rates of political proselytizing.

5. Accounting for the impact of female candidates
Why, exactly, should a candidate's gender affect the electoral context, and lead to higher rates of political proselytizing by either men or women? And why was 1992 different? Several hypotheses may be suggested:

1. The media effect: A woman candidate for major office is still something of a novelty. The media pay more attention to such races, and there was considerable hype about the "year of the women" in both 1990 and 1992 (Wilcox 1994:2), including cover stories in mass-market women's magazines such as McCall's and Redbook. Greater media coverage of women's candidacies and issues should enhance media consumption, especially among women (9). In 1994, however, other issues dominated, and the media paid less attention to female candidates (Berke 1994a).

2. The information effect. Issue awareness is dismally low in most Congressional elections. Busy voters, with many concerns other than politics, often rely on heuristics (party, their own issue positions) to reduce information costs (Mondak 1995; Conover and Feldman 1989). Women candidates, since they are unusual, are more likely to be noticed and remembered by voters (perhaps in part because of heightened media coverage.) I therefore anticipate greater name recognition for female candidates (10). Women should also be more aware of the issues in states with women on the ballot for major office.

3. Political involvement. The presence of women on the ballot should heighten political interest and involvement in the campaign, especially for women (as (Koch 1997) observed for 1992.) This may of course reflect greater media coverage and greater knowledge of the candidates and/or issues. Symbolic representation (as discussed above) may also be empowering for women as it is for African-Americans and Latinos,

4. The efficacy (or self-esteem) effect. Women's political socialization does not often stress an active public role for women. As good "subjects," women may dutifully vote, but often feel themselves less certain of political issues, less interested, and less well informed (Sapiro 1983). Certainly women rank lower in terms of political efficacy (although as feminist scholars have warned, political scientists should avoid using male behavior as the norm: Bourque and Grossholtz, 1974). But the presence of competent, experienced women "role models" running for major office may change these perceptions (Mandel, 1993: 55). Citizens confident of their own political knowledge and competence should be more likely to attempt to influence others' views.

5. The impact of party, a major filter through which political information is perceived. We would expect political involvement to be greater if the woman running in one's state or district is from one's preferred party.

6. Issue emphasis. Women office-holders stress somewhat different issues than their male counterparts, including several of particular concern to women: education, family leave, child care, abortion rights (Carroll et al., 1991; Thomas, 1994). One need not assume that all women share feminist values, or support female candidates regardless of her stand on specific issues. Neither should one expect all men to be opposed to feminism, or to women candidates; in 1990 Nancy Landon Kassebaum garnered more votes from men than from women (Cook, 1994). "Woman" is a broad category encompassing a range of interests (Sapiro 1981; Young 1994). However, persons concerned about those interests (whether they agree or disagree on the options) should be more likely to use their political voice if women are running or gender-related issues are salient. In 1992, three issues in particular served to make gender-related concerns highly salient: abortion, sexual harassment, and the underrepresentation of women in the Senate. The Clarence Thomas-Anita Hill controversy focused public and media attention on these last two issues, and we would expect these issues to have a greater impact on women than on men.

[Table 4 about here]

Despite the "Year of the Woman" hype, most women in 1992 behaved much as they had in other election years (Table 4). Women's interest in politics generally and in the 1992 election campaign, although higher than in any election year since 1972 (11), was still lower than men's, and their overall media use was similar. Women were less likely than men to correctly identify the names or the parties of candidates running for House or Senate (the "knowledge of candidates" measure), and fewer women could locate themselves on the NES seven-point issue-position scales (the "knowledge of issues" measure.) But they did show more interest than men did in "news about issues that especially affect women." Men and women in 1992 reported listening to TV or radio talk shows at exactly the same rate, 55 percent (unfortunately the NES did not ask respondents if they were callers as well as listeners.)

Marked gender differences were also evident in what Niemi et al. (1991) have termed "internal political efficacy" (the term "political self-esteem" is used here to distinguish this measure from the NES's original scale of internal political efficacy, which was based on different questions). The Niemi et al. measure is based on a series of questions about respondents' sense of their own political competence and understanding, asked by the NES for the first time in 1988 and repeated in 1992 (but not in 1990 or 1994). Only 25 percent of women, compared with 39 percent of men, scored high on the overall index created from these questions, and college-educated women scored about the same as men with less than a high school education (12).

As Table 4 shows, however, the presence of women on the ballot in 1992 (for all offices combined) is associated with higher levels of political self-esteem and of the other indicators of media use, political knowledge, and involvement in politics, for both men and women. The differences are not large in percentage terms, but almost all the gamma coefficients (a standard measure of association for ordinal data) are positive and most are statistically significant at the usual .05 level. The most striking impact of female candidates is on women's knowledge of issues, their interest in the campaign, their use of media, and their propensity to discuss politics with family and friends. In most cases, the relationship between these indicators and the presence of female candidates is stronger for women than for men. One exception is that men`s candidate recognition was higher in the presence of a female candidate.

As Table 4 also shows, women as a group are considerably more feminist than men (except that more men than women believe women need to work together, rather than individually.) But when it comes to attitudes toward feminism, the presence of female candidates appears to be modestly positive for women, but negative for men. The presence of a female candidate might thus stimulate men's knowledge and awareness, but could also engender opposition; this possibility will be explored further with multivariate techniques.

Let us now consider whether people concerned with particular issues, or issue positions, were more likely to engage in political proselytizing in 1992. Greater concern with a number of so-called "women's issues" (abortion, child care, family leave, social services, education, sexual harassment) could motivate more women to attempt to influence others' votes. The NES survey, unfortunately, does not ask respondents WHY they attempt political persuasion, or why they supported or opposed women candidates (13). But based on the long-standard question "What do you think are the most important issues facing the country?" there is little difference between men and women who proselytize; both sexes are most concerned with economic issues (jobs, the deficit, the economy.) Women were slightly more likely to mention social issues (poverty, the homeless); men were more concerned with health care, education, and the environment. Very few respondents cited defense, foreign affairs, or general governmental issues; those who did tended to be male. But the overall rankings of issue priorities are similar for proselytizers of both sexes.

Conspicuous by their absence as "most important problems" are issues usually identified as prominent on the feminist agenda. Of the approximately seventy abortion-related responses to the "most important national problems" questions in 1992, only 9 articulated a pro-choice position. Pro-life supporters were much more likely to attempt to influence others' votes. Similarly, women's issues such as child care, equal pay, sexual harassment, or family leave attracted only a couple of responses each. Since male political proselytizers were especially concerned with the deficit, it is not surprising that they were disproportionately Perot supporters. In short, there is little evidence from the open-ended NES question on national problems that a particular agenda (feminist or not) motivated the women who engaged in political discussion in 1992.

When we consider responses to closed-ended questions on issues with particular relevance for women, however, a different picture emerges (Table 5). Four issues (abortion, sexual harassment, opinions on Hill versus Thomas, equality for women) were identified by Paolino (1995) as particularly relevant for votes for female Senate candidates in 1992 (14). On three of these issues men holding the more conservative position are more likely to attempt to sway others' votes, while women holding liberal or feminist positions are the more likely proselytizers (although by narrower margins.) But both men and women who support equal roles for women are more likely to attempt to influence others' votes than are those who do not. Considerable gender differences are evident in responses to all four issues, as the N's illustrate; female respondents are consistently more feminist than men. Given the sharp differences in outlook between male and female proselytizers, one might also expect some acrimonious debates, since those holding the polar positions are more likely than those in the middle to attempt to sway others' votes (15).

[ Table 5 about here ]

Table 5 also shows the impact of party on proselytizing in the context of female candidates for Senate. Given the paucity of Republican women running in 1992, we cannot fully test for the impact of party on political involvement (the NES conducted no interviews in South Dakota where Charlene Haar was the Republican Senatorial candidate). But in the states in which Democratic women were on the ballot for Senate, rates of political proselytizing by BOTH Republican and Democratic women were higher than elsewhere. And there was little difference between the parties; 58 percent of both Republican women and Democratic women reported attempts to influence others' votes in California. For men in 1992, the party pattern was reversed: Republican men were slightly less likely to discuss politics if a woman was on the ballot for Senate, Democratic men significantly less so.

As is clear from Table 5, women by no means agree on these issues; party, age, race, religious affiliation, and education affect their views (see Alvarez and Brehm 1995, for discussion of the complexity of opinions on abortion.) Nevertheless, Republican women proselytizers are more liberal and feminist than Republican men with respect to opinions on abortion, equal roles, and sexual harassment. There is thus an affinity between women political persuaders in 1992 and the predominately liberal and Democratic women who ran for the House, Senate, or governorships. We lack direct evidence that such issue preferences encouraged political discourse, and there were too few Republican women candidates in the NES sample to fully test the impact of party (16). But the possibility remains that candidate gender functions as a useful shorthand or heuristic for voters by reducing information costs. Both men and women could reasonably assume that a Democratic woman running for the U.S. Senate would be pro-choice, liberal, and generally sympathetic to women's concerns, even though some women (particularly Dianne Feinstein and Carol Moseley-Braun) deliberately downplayed these qualities in 1992 in order to broaden their base of support (17).
 

6. Testing for the impact of candidate gender
Evidence presented thus far suggests that a campaign-specific factor (candidate gender) contributed to political proselytizing in 1992, and did so in part because of its impact on media use, self-esteem, political awareness and involvement, and issues relevant to women. But do campaign-specific factors have an independent effect on men's or women's propensity to influence others' votes once other factors are controlled for? I again make use of logistic regression, adding issues and factors specific to the 1992 campaign to the baseline model estimated in Table 3. Since men and women appear to have responded quite differently to the presence of female candidates, we will estimate separate equations for each sex. Because the presence of female candidates appears to increase issue awareness, we will also substitute our measure of knowledge of issues for the education scale used earlier: because of multicollinearity, they cannot be used in the same equation (18). We will also factor in support for Perot, whose controversial on again-off again candidacy in 1992 stimulated considerable political discussion (Alvarez and Nagler, 1995); his supporters were more likely than Clinton or Bush supporters to attempt to influence others' votes (48 compared with 42 and 46 percent proselytizers, respectively.)

The models estimated initially included the abortion question as well as the Hill-Thomas difference score, the question on salience of sexual harassment, and two different indicators of support for equality for women. But the latter three measures, as well as the scale of media use, did not meet minimum criteria for inclusion in the equations, either separately or in combination (19). The results are shown in Table 6, with separate equations for all female candidates and for female Senate candidates. As Table 3 suggested, the presence of female candidates encourages political proselytizing by women, and discourages it for men (most strikingly in the Senate-only equations.) The coefficients for self-esteem and knowledge of issues are significant as well, but female candidates appear to have an impact over and above their contribution to higher levels of self-esteem, media use, or issue awareness (20). Candidate gender may thus function for women to suggest symbolic representation of their interests. It may also serve as a useful cue for "bounded rational" voters who may be unable to remember candidates' names, issue positions, or characteristics (Lodge et al. 1995), but who do recall the sex of the candidates.

[ Table 6 about here ]

The Perot factor is positive (although not significant) for both sexes, even though men were disproportionately Perot supporters (23 percent of men voted for him, but only 14 percent of women.) Our contextual interpretation of political proselytizing must therefore consider Perot's candidacy as well as the impact of women running in 1992.

The issue coefficients are surprising. Despite the salience of the Hill-Thomas hearings and the sexual harassment issue, the Hill-Thomas scale is statistically significant, but of marginal importance, only for women in the equations for the Senate candidates (21). And all of the coefficients for the abortion scale are negative, for both men and women (although more so for men.) Once other factors are controlled, those with pro-life attitudes are more likely to proselytize. As noted earlier, the abortion responses to the open-ended questions on national problems came predominately from those holding pro-life positions; Verba et al. (1995: 261) note that pro-lifers often focus their participatory activities on the abortion issue. Overall, however, these more detailed equations confirm what was suggested by Table 3. While political interest, partisanship, and age still account for much of the variance in political proselytizing, 1992 was a unique election because of the presence of female candidates, the focus on issues of concern to women, and the positive impact of female candidates on women's attempts to influence others' votes. The addition of these campaign-specific factors significantly improves the predictions.

7. Conclusion
Political proselytizing affords both theoretical and methodological advantages for the analysis of gender differences in political participation. Unlike many other forms of conventional political participation, the gender gap in political persuasion has persisted since the 1970's. It has been well established that social resources and enduring political attitudes help determine whether or not one engages in political activity. But using a contextual approach, looking at gender differences over time and in varying electoral situations, permits us to move beyond individual-level analysis of participation. How men and women define their own political roles depends in part on the choices available to them. This confirms feminist analyses (Tannen 1993; Crawford 1995) showing that the impact of gender on discourse depends on the social context.

Political proselytizing varied systematically in 1992 according to an important contextual variable: the presence of women as major-party candidates for office. As in 1990, this effect held for both men and women, but was stronger for women in 1992, and was most evident where women ran for the Senate. In 1992 as well as in 1972, men remain significantly more likely than women to report attempts to influence others' votes. But the size of this difference has varied considerably across time. It is larger in Presidential than in off-year elections; it was smallest in 1984 when Geraldine Ferraro was on the ballot. And at least in 1992, the "gender gap" in political persuasion disappeared altogether in states with more than one woman running for office; in both 1990 and 1992 the gap was smaller when a woman ran for the House or for Governor. Further, in both 1990 and 1992, the presence of a woman on a major-party ballot was linked to increased political awareness and more attempts at political persuasion by women.

But why, as the logistic regressions showed, were men's odds of attempting to influence others' voters lower when women appeared on the ballot? As MacKuen's (1990) analysis suggested, some may have chosen to "clam" rather than to risk arguments with women, or with persons of either sex holding more feminist issue positions. Given the overrepresentation of conservative males on talk shows, such reticence is unexpected. As Cook (1994:224) found, men were significantly less likely than women to vote for women running for the Senate, regardless of party (although with considerable differences from state to state). A possible explanation is the declining support for feminism and the women's movement among white men since 1990 (based on NES feeling thermometer trends; see Hansen 1996.) But focus groups or some other form of in-depth interviewing may be necessary to explore these relationships. Nevertheless, as Table 1 indicated, men's political proselytizing was higher in 1992 than in any Presidential election since 1976. The "Perot factor" is one possibility, because Perot supporters were disproportionately male. Thus political context appears to matter for men as well as women.

The importance of context is underscored by the much weaker impact of female candidates on political proselytizing in 1994. In part, this reflects the very different stimulus provided by midterm elections, when many fewer women (and men) attempt to influence others' votes, but issues and candidates were also different in 1994. Although a sizeable number of women ran for office (CAWP 1994), many more were Republican and/or conservative than in 1992, and many of these successfully defeated the more liberal and Democratic women elected in 1992 (Marshall 1995). Women were still more liberal than men in 1994; in fact, the gender gap in voting for Congress was larger than in the previous eight elections (Berke 1994b). But women as well as men voted in a more conservative vein than in 1992, and turnout among Republicans was high (22). The issues shifted as well, with attention focused on the budget, crime, and welfare reform rather than on sexual harassment or the representation of women (Berke 1994a). The 1994 pattern suggests that it is not the presence of women alone, but women focusing on issues of concern to women, that mobilizes women to proselytize. But we will need to examine future Presidential elections in order to determine whether 1992 was truly unique or presaged a different pattern for women's political involvement.

These results present a much different picture of political discussion in American politics than in some recent studies (23). Although the "gender gap" in political proselytizing has persisted from the 1970's to the 1990's, political persuasion can be a valid option for women when female candidates appear to engage their interests or concerns. The same may be true for Independents when a third-party candidate appears. In many states --especially California, where female proselytizers outnumbered male in 1992--American women were not hesitant to express political views, even at some risk of politicizing social relationships or creating conflict by attempting to influence others' votes. And higher scores on the self-esteem scale suggest that female candidates enhanced women's confidence in their own abilities to act politically. Women, especially younger and better-educated women, are thus behaving more as citizens than as subjects by taking part in public political discourse and by attempting to exert political influence.

These findings thus reinforce Carroll and Zerilli`s summation of research on women and politics since 1970: "Differences between women and men, whether in elites or mass publics, were viewed as socially constructed and thus changeable rather than as natural and immutable" (1993:62). The systematic variation in women's political response across time and elections further suggests that women are not limited by their social status or socialization to private, subservient or nonconflictual political roles, but are proactive citizens who can choose when and how to use their voices.
 

--------------------------------------------------------------------------------

APPENDIX

Coding of variables used in the analysis:

POLITICAL PROSELYTIZING: "During the campaign, did you talk to any people and try to show them why they should vote for [1984-1994 or against] one of the parties or candidates?" 1=no, 2=yes.

DISCUSS POLITICS: "Do you ever discuss politics with your family or friends?" 1=no, 2-yes.

FEMALE CANDIDATES: every state or Congressional district with a female Democratic or Republican candidate for Senate, House, or Governor was coded 1, otherwise zero. For logistic regression, the number of female major-party candidates running in a respondent's district was coded as none, one, or two or more. Data from CQ Weekly Report election issues (November 1990, 1992, 1994) and CAWP annual fact sheets on elected women.

CARE ABOUT ELECTION: "Generally speaking, would you say that you personally care a great deal who wins the presidential [Congressional in off years] election this fall, or that you don`t care very much?"

FOLLOW PUBLIC AFFAIRS: "Would you say that you follow what's going on in government and public affairs most of the time, some of the time, only now and then, or hardly at all? "

CAMPAIGN INTEREST: "Some people don't pay much attention to political campaigns. How about you? Would you say you are very much interested, somewhat interested, or not much interested in the political campaigns so far this year?"

POLITICAL INTEREST: linear combination of the CAMPAIGN INTEREST and FOLLOW POLITICS responses, following Bennett (1986). Recoded to range from 2 (low on both) to 7 (high on both)

POLITICAL SELF-ESTEEM: Using Niemi et al.'s (1991) questions on internal political efficacy, an additive scale was constructed scoring (inverted) answers of 0 to 4 (strongly disagree to strongly agree) for each question (1992 only):

"I feel that I have a pretty good understanding of the important political issues facing our country."

"I consider myself well qualified to participate in politics."

"I feel I could do as good a job in public office as most other people."

"I think that I am better informed about politics and government than most people."

Range 0 to 16; Cronbach's Alpha of .78.

PARTISANSHIP: Strength of party identification, based on NES seven-point scale. Strong Democrats or Republicans coded 3, weak or leaning 2, independents 1, apolitical 0.

RACE: Whites coded 1, Blacks 2 (others omitted)

SEX: Males coded 0, females 1

EDUCATION: NES 7-point scale, ranging from grade school to postgraduate.

MEDIA: reliance on media sources used during the last campaign. Additive scale, 5 points for each response "a great deal," 4 for "quite a bit," 3 for "some," 2 for "very little," and 1 for "none" for newspapers, magazines, and television; for radio, 5 points listening to a "good many" speeches or discussions about the campaign, 3 points for "several," 1 point for "one or two."

TALK SHOWS: NES variable 3210, "Do you listen to or watch talk shows on radio or TV where people call in to voice their opinions?" 0=no, 1=yes.

KNOWLEDGE OF CANDIDATES: all those correctly identifying both candidate and party for a given office (House or Senate) are coded 1, otherwise 0

KNOWLEDGE OF ISSUES: additive scale based on responses to NES 7-point self-positioning scales on defense spending, job creation, health care, and spending on government services. Respondents received one point for each response in the 1-7 range for each scale, and 0 if they could not so place themselves. Range from 0 to 4.

PEROT: dummy variable coded 1 if respondent reported intending to vote for Perot (pre-election survey), 0 otherwise.

ABORTION: NES variable 3732 giving four options for the legal status of abortion, ranging from 1, "By law, abortion should never be permitted" to 4, "By law, a woman should always be able to obtain an abortion as a matter of personal choice."

SEXUAL HARASSMENT: NES variable 3741: "How serious a problem do you think sexual harassment in the work place is?" Recoded as 1=not too serious, 2=somewhat serious, 3=very serious

HILL-THOMAS SCALE: difference between feeling-thermometer degrees for Anita Hill and Clarence Thomas.

POWER/DISCONTENT: following Paolino (1995: 312), scale ranging from 0 (women have more power but men should) to 5 (men have more power but women should), combining NES variables 6007 and 6008.

EQUAL ROLES FOR WOMEN: NES 7-point self-anchoring scale, ranging from 1, "Men and women should be equal," to 7, "woman's place is in the home."

WEIGHTING: following ICPSR guidelines, 1992 data were weighted using V7000 for comparability with earlier, unweighted studies, and 1994 data were similarly weighted using V6, to compensate for attrition among panel respondents.
 
 

--------------------------------------------------------------------------------

NOTES

Data for this study were provided by the Inter-University Consortium for Political Research. Thanks to my graduate research assistants Simeon Brodsky, Carolyn Dudek, Vincent McIlhinney, and Michael Kulischek for help with data collection and analysis.

1. A recent study by the American Association of University Women (Bailey et al. 1992) also found that young women were less likely to speak up in class and less likely to be called on by teachers. The study documents many ways in which the educational experience devalues women's speech.

2. The 1992 NES contains a broader range of questions on political discussion, but not all of these were asked in earlier years. See analysis by Bennett et al. (1994). The question on political proselytizing shows only a modest relationship (R=.41) with a question on general discussion of politics with family and friends.

3. This decline in political discussion in Congressional years may reflect the increased use of negative campaign messages. As Ansolabehere et al. (1994) demonstrate, negative campaigns turn off voters and decrease turnout, and may inhibit political discussion as well. As will be shown below, in 1992 this trend may have been counterbalanced by the excitement and controversy stirred up by Ross Perot and by the number of female candidates.

4. The NES 1988-1990-1992 panel study of US Senate races includes more states, but unfortunately did not include the question on political proselytizing. Huckfeldt and Sprague's (1995) snowball sample could identify the persons with whom respondents discussed politics and could assess the effects of attempts at influence, but they could not test for variations in electoral cues based on their one-community study.

5. If a woman was the incumbent, over 90 percent of both males and females could identify her correctly, compared with less than 70 percent in other elections. Although there were a handful of House races where women opposed each other, too few NES respondents from those districts were interviewed in any of these three elections to permit a further test of the impact of women opposing each other. The same problem of lack of respondents from Maryland or Kansas precluded any analysis of the impact of the female Senate incumbents (Barbara Mikulski in 1992, Nancy Landon Kassebaum in 1990).

6. A few states in the NES 1992 study had even higher rates of political proselytizing by women than did California, but in none of these were more than a handful of interviews conducted. We will therefore consider the general impact of women in the ballot rather than the effect of individual women's candidacies. See Cook (1994: 220-225) for an attempt to sort out the relative impact of party and gender in the 1992 Senate races; as she notes, turnout among BOTH men and women was higher in races with women on the ballot.

7. See Darcy et al. (1994) for a discussion of the political and institutional factors (incumbency, legislative professionalism, single-member districts, party structure, political culture) that account for differences in the proportion of women elected to state legislatures.

8. In 1992 and 1994, younger voters were considerably more likely than older ones (those over 60) to attempt to influence others' votes. The gender gap persists across all age groups in all years, and is largest among older women.

9. In the 1980's, media coverage appeared to be detrimental to women candidates (Kahn and Goldenberg 1991), but Williams (1994) found that women candidates in 1992 benefitted from savvy use of advertising content and style to emphasize both their femininity and their competence. Kahn's (1992) experiments suggest that women tend to view female candidates more favorably regardless of the content of media coverage.

10. See Lodge et al. (1995) for an experimental assessment of the weakness of recall measures in candidate evaluation.

11. The "political interest" index (following Bennett 1986) showed that women's interest in politics and in the campaign was higher in 1992 than in any previous election, but still significantly lower than men's. In 1994, male interest remained high, but women's was about average for the period since 1972 (Hansen 1996).

12. In 1992, a significant number of men were confident of their own political knowledge even if their objective levels of education or political knowledge was low, while women's self-esteem was more closely linked to political knowledge and level of education (Hansen 1996).

13. See Schlozman et al. (1995) for a sophisticated linking of issue priorities and types of participation.

14. The equal-role seven-point scale is used here in preference to Paolino's power/discontent scale, since the latter has a highly skewed distribution (the vast majority of NES respondents believed that men and women are equal and should be.)

15. Contrary to Noelle-Neumann`s (1993) findings that people prefer to discuss politics with those of a similar bent, the Times Mirror (1993) study reported that talk-show listeners were happiest when an opinion DIFFERENT than their own was being expressed.

16. One can only speculate as to the content of Republican women's discourse, but we do know that many such women crossed party lines to support a female candidate in 1992. In 1990, by contract, relatively few Democratic women supported the Republican women running for Senate, all but one of whom lost (Cook 1994: 220). The CAWP report on the 1994 election found significant (at least four percent) gender gaps in 81 percent of Congressional elections for which exit-poll data were available (1994: 7), and suggested that women crossed party lines in several of these races to support women candidates.

17. I wanted to explore the intersection of race and gender with reference to the election of Carol Moseley-Braun in Illinois, but the NES sample included only eight African-American women in that state. As Jelen's (1994) analysis suggests, however, Moseley-Braun's candidacy did attract strong support in the Black community, especially among women. Although she downplayed racial issues sufficiently to attract votes from white Democrats Downstate, she garnered 95 percent of the Black vote and 58 percent of women's votes. See also Stanbach (1988-89) on black women's talk.

18. In logistic regression, the presence of multicollinearity is suggested by high estimated standard errors when two related variables are included in the same equation (Hosmer and Lemeshow 1990:132).

19. Media use also showed strong indications of multicollinearity with both the education and knowledge variables, and is therefore excluded from Table 6.

20. Equations were also run using interaction terms to test whether the presence of female candidates strengthened the impact of knowledge, self-esteem, media use, or political interest upon political proselytizing. Although the (estimated) coefficients for the interaction terms were positive, all were too small to meet minimum criteria for inclusion in the equation. The Wald statistic has a Chi-square distribution and tests whether a logistic regression coefficient differs significantly from zero (Norusis 1990:48).

21. The 1992 NES included several other questions on sexual harassment, including one on the importance of the issue (see Table 5), but none of these performed any better than the Hill-Thomas difference score in logistic regressions predicting political proselytizing.

22. As Mandel noted (cited in CAWP 1994:5), "the electorate's energy [in 1994] was focused less on empowering newcomers than on ejecting old timers" -- including several of the Democratic women elected in 1992.

23. Bennett et al. (1994) did NOT find gender to be a significant predictor of frequency of engaging in political discussion in 1992, but they did not include the contextual variable for women candidates.

REFERENCES

Abramovitz, Alan I. 1995. "It's Abortion, Stupid: Policy Voting in the 1992 Presidential Election." Journal of Politics 57:176-186.

Aldrich, John, and Forrest Nelson. 1984. Linear Probability, Logit, and Probit Models. Beverly Hills, CA: Sage.

Alvarez, R. Michael, and Jonathan Nagler. 1995. "Economics, Issues, and the Perot Candidacy: Voter Choice in the 1992 Presidential Elections." American Journal of Political Science 39:714-744.

Alvarez, R. Michael, and John Brehm. 1995. "American Ambivalence toward Abortion Policy." American Journal of Political Science 39: 1055-1079.

Ansolabehere, Stephen, et al. 1994. "Does Attack Advertising Demobilize the Electorate?" American Political Science Review 88:829-840.

Bailey, Susan M., et al. 1992. The AAUW Report: How Schools Shortchange Girls. Washington, D. C.: National Education Association.

Belenky, Mary Field, et al. 1986. Women's Ways of Knowing: The Development of Self, Voice, and Mind. New York: Basic Books.

Bennett, Stephen Earl. 1986. Apathy in America 1960-1984: Causes and Consequences of Citizen Political Indifference. Dobbs Ferry, NY: Transnational Publishers.

Bennett, Linda L., and Stephen E. Bennett. 1989. "Enduring Gender Differences in Political Behavior; the Impact of Socialization and Political dispositions." American Politics Quarterly 17:105-122.

Bennett, Stephen E., Bonnie Fisher, and David Resnick. 1994. "Speaking of Politics in the United States." Paper presented at the Midwest Political Science Association, Chicago, April.

Berelson, Bernard, Paul Lazarsfeld, and William McPhee. 1954. Voting. Chicago: University of Chicago Press.

Berke, Richard L. 1994a. "In '94, 'Vote for Woman' Does Not Play So Well." New York Times, October 3, A1, A8.

Berke, Richard L. 1994b. "Victories were Captured by GOP Candidates, not the Party's Platform." New York Times, Nov. 10, B1, B4.

Bobo, Lawrence, and Frank D. Gilliam. 1990. "Race, Sociopolitical Participation, and Black Empowerment." American Political Science Review 84:377-393.

Bookman, Ann, and Sandra Morgen. 1988. Women and the Politics of Empowerment. Philadelphia: Temple University Press.

Books, John, and Charles Prysby. 1988. "Studying Contextual Effects on political behavior." American Politics Quarterly 16:211-238.

Bourque, Susan C., and Jean Grossholtz. 1974. "Politics an Unnatural Practice; Political Science looks at Female Participation." Politics and Society 4:225-266.

Brady, Henry E., Sidney Verba, and Kay Lehman Schlozman. 1995. "Beyond SES: a Resource Model of Political Participation." American Political Science Review 89:271-294.

Carroll, Susan J., et al. 1991. The Impact of Women in Public Office: An Overview. New Brunswick: NJ: Center for American Women in Politics.

Carroll, Susan J., and Linda M. G. Zerilli. 1993. "Feminist Challenges to Political Science." In Ada W. Finifter, ed., Political Science: The State of the Discipline II. Washington, D. C.: APSA.

Center for American Women and Politics (CAWP). 1994. "1994 Post-Election Wrap-Up." New Brunswick, NJ: Eagleton Institute.

Clark, Cal, and Janet Clark. 1986. "Models of Gender and Political Participation." Women and Politics 6:5-25.

Conover, Pamela, and Stanley Feldman. 1989. "Candidate Perception in an Ambiguous World; Campaigns, Cues, and Inference Processes." American Journal of Political Science 33:912-940.

Cook, Elizabeth A., and Clyde Wilcox. 1991. "Feminism and the Gender Gap - a Second Look." Journal of Politics 53:1111-1122.

Cook, Elizabeth A. 1994. "Voter Response to the Women Senate Candidates." In Elizabeth A. Cook, Sue Thomas, and Clyde Wilcox, eds., The Year of the Woman: Myths & Realities. Boulder: Westview

Crawford, Mary. 1995. Talking Difference: On Gender and Language. London: Sage

Darcy, R., Susan Welch, and Janet Clark. 1994. Women, Elections, and Representation, 2nd ed. Lincoln: University of Nebraska Press.

Elshtain, Jean Bethke. 1981. Public Man, Private Woman: Women in Social and Political Thought. Princeton: Princeton University Press.

Elshtain, Jean Bethke. 1982. "Feminist discourse and its Discontents: Language, Power, and Meaning." Signs 7:603-621.

Elshtain, Jean Bethke. 1989. "Issues and Themes in the 1988 Campaign." In Michael Nelson, ed., The Elections of 1988. Washington: CQ Press

Ferejohn, John A. 1990. "Information and the Electoral Process." In John A. Ferejohn and James H. Kuklinski, eds., Information and Democratic Processes. Urbana: University of Illinois Press.

Fowlkes, Diane. 1992. White Political Women: Paths from Privilege to Empowerment. Knoxville: University of Tennessee Press.

Gilligan, Carol. 1982. In a Different Voice: Psychological Theory and Women's Development. Cambridge: Harvard University Press.

Goldin, Claudia. 1990. Understanding the Gender Gap: The Economic History of American Women. New York: Oxford University Press.

Hagle, Timothy M., and Glenn E. Mitchell II. 1992. "Goodness of Fit Measures for Probit and Logit." American Journal of Political Science 36:762-784.

Hansen, Susan B., Linda M. Franz, and Margaret Netemeyer-Mays. 1976. "Women's Political Participation and Policy Preferences." Social Science Quarterly 54:576-590.

Hansen, Susan B. 1996. "Political Proselytizing and the Social Construction of Gender, 1972-1994." Paper presented at the Midwest Political Science Association, Chicago, April.

Hero, Rodney E. 1992. Latinos and the U. S. Political System: Two-Tiered Pluralism. Philadelphia: Temple University Press.

Hosmer, David W., Jr., and Stanley Lemeshow. 1990. Applied Logistic Regression. New York: John Wiley & Sons.

Houston, Marsha, and Cheris Kramarae. 1991. "Speaking from Silence: Methods of Silencing and of Resistance." Discourse and Society 2:387-399.

Huckfeldt, Robert, and John Sprague. 1991. "Discussant Effects on Vote Choice: Intimacy, Structure, and Interdependence." Journal of Politics 53:122-158.

Huckfeldt, Robert, and John Sprague. Citizens, Politics, and Social Communication. New York: Cambridge University Press, 1995.

Jelen, Ted G. 1992. "Political Christianity: A Contextual Analysis." American Journal of Political Science 36:692-712.

Jelen, Ted G. 1994. "Carol Moseley-Braun: The Insider as Insurgent." In Elizabeth A. Cook, Sue Thomas, and Clyde Wilcox, eds., The Year of the Woman: Myths & Realities. Boulder: Westview.

Kahn, Kim Fridkin. 1992. "Does Being Male Help? An Investigation of the Effects of Candidate Gender and Campaign Coverage on Evaluations of U.S. Senate Candidates." Journal of Politics 54:497-517.

Kahn, Kim Fridkin, and Edie N. Goldenberg. 1991. "Women Candidates in the News." Public Opinion Quarterly 55:180-191.

Kanter, Rosabeth Moss. 1977. Men and Women of the Corporation. New York: Basic Books

Kathlene, Lyn. 1989. "Uncovering the Political Impacts of Gender: an Exploratory Study." Western Political Quarterly 42:397-421.

Kathlene, Lyn. 1994. "Power and Influence in State Legislatures; the Interaction of Gender and Position in Committee Hearing Debates." American Political Science Review 88:560-576.

Kathlene, Lyn. 1995. 'Alternative Views of Crime; Legislative Policy-Making in Gendered Terms." Journal of Politics 57:696-723.

Katz, Elihu, and J. J. Feldman. 1962. "The Debates in Light of Research: a Survey of Surveys." In Stanley Kraus, ed., The Great Debates. Bloomington, IN: Indiana University Press.

Keeter, Scott. 1985. "Public opinion in 1984." In Gerald M. Pomper et al., The Election of 1984: Reports and Interpretations. Chatham, NJ: Chatham House.

Kinder, Donald R., and Don Herzog. 1993. "Democratic Discussion." In George E. Marcus and Russell L. Hanson, eds., Reconsidering the Democratic Public. University Park: Penn State University Press.

Koch, Jeffrey. 1997. "Women's Engagement in Politics in the Year of the Woman." American Politics Quarterly, forthcoming.

Kramarae, Cheris. 1982. "Gender: How She Speaks." In Ellen B. Ryan and Howard Giles, eds., Attitudes Toward Language Variation. London: Edward Arnold.

Lodge, Milton, Marco R. Steenbergen, and Shawn Brau. 1995. "The responsive Voter: Campaign Information and the Dynamics of Candidate Evaluation." American Political Science Review 89:309-326.

MacKuen, Michael. 1990. "Speaking of Politics: Individual Conversational Choice, Public Opinion, and the Prospects for Deliberative Democracy." In John A. Ferejohn and James H. Kuklinski, eds, Information and Democratic Processes. Urbana: University of Illinois Press.

Mandel, Ruth. 1993. "The Political Woman." In Sherri Matteo, ed., American Women in the Nineties. Boston: Northeastern University Press.

Margolis, Jane. 1992. "Piranhas, Monsters, and Jugglers: the Psychology of Gender and Academic Discourse." On Teaching and Learning 4:5-26.

Marshall, Marcia A. 1995. "Women of the Class of '92: Winners and Losers in the 1994 House Election." Paper presented at the American Political Science Association, Chicago, September.

Miller, Arthur. 1988. "Gender and the Vote, 1984." In Carol Mueller, ed., The Politics of the Gender Gap. Newbury Park, CA: Sage.

Miller, Warren E., and the National Election Studies. 1994. American National Election Studies Cumulative Data File, 1952-1992, 7th release. Ann Arbor, MI: University of Michigan, Center for Political Studies.

Mondak, Jeffrey. 1995. "Media Exposure and Political Discussion in U. S. Elections." Journal of Politics 57:62-85.

Mueller, Carol, ed. 1988. The Politics of the Gender Gap. Newbury Park, CA: Sage.

Mueller, Carol. 1991. "The Gender Gap and Women's Political Influence." In Janet K. Boles, ed., American Feminism: Issues for a New Movement. Annals of the American Academy of Political and Social Science 515:23-37.

Munson, Wayne. 1993. All Talk; Talk Shows in Media Culture. Philadelphia: Temple University Press.

Niemi, Richard G., Stephen C. Craig and Franco Mattei. 1991. "Measuring Internal Political Efficacy in the 1988 National Election Study." American Political Science Review 85:1407-1416.

Noelle-Neumann, Elisabeth. 1993. The Spiral of Silence: Public Opinion--Our Social Skin, 2nd ed. Chicago: University of Chicago Press.

Norusis, Marija J. 1990. SPSS Advanced Statistics User's Guide. Chicago: SPSS Inc.

Paolino, Philip. 1995. "Group-Salient Issues and Group Representation: Support for Women Candidates in the 1992 Senate elections." American Journal of Political Science 39:294-313.

Rapoport, Ronald B. 1981. "The Sex Gap in Political Persuading: where the `Structuring Principle' works." American Journal of Political Science 25:32-48.

Reskin, Barbara. 1993. "Sex Segregation in the Workplace." Annual Review of Sociology 19:241-270.

Robinson, John. 1976. "Interpersonal Influences in Election Campaigns: Two Step-Flow Hypotheses." Public Opinion Quarterly 40:304-319.

Rosenstone, Steven J., and John Mark Hansen. 1993. Mobilization, Participation, and Democracy in America. New York: MacMillan.

Rosenstone, Steven J., Warren E. Miller, Donald R. Kinder, and the National Election Studies. 1995. American National Election Study, 1994. Ann Arbor, MI: University of Michigan, Center for Political Studies.

Sapiro, Virginia. 1981. "Research Frontier Essay: When are Interests Interesting? The Problem of Political Representation of Women." American Political Science Review 75:701-716.

Sapiro, Virginia. 1983. The Political Integration of Women. Urbana, IL: University of Illinois Press.

Schlozman, Kay Lehman, et al. 1995. "Gender and Citizen Participation: Is There a Different Voice?" American Journal of Political Science 39:267-286.

Schlozman, Kay Lehman, Nancy Burns, and Sidney Verba. 1994. "Gender and the Pathways to Participation: the Role of Resources." Journal of Politics 56:963-990.

Smithey, Shannon, and Jon Hurwitz. 1996. "Gender Differences toward Crime and Punishment." Paper presented at the Midwest Political Science Association, Chicago, April.

Stanbach, M. 1988-89. "Feminist Theory and Black Women's Talk." Howard Journal of Communications 1:187-194 .

Steinem, Gloria. 1986. "A Woman Runs for Vice-President." In James David Barber and Barbara Kellerman, eds, Women Leaders in American Politics. Englewood Cliffs, NJ: Prentice-Hall.

Tamerius, Karin. 1993. "Does Sex Matter? Women Representing Women's Interests in Congress." Paper presented at the Midwest Political Science Association, April.

Tannen, Deborah. 1994. Gender and Discourse. New York: Oxford

Tannen, Deborah, ed., 1993. Gender and Conversational Interaction. New York: Oxford University Press.

Tate, Katherine. 1991. "Black Political Participation in the 1984 and 1988 Presidential Elections." American Political Science Review 85:1159-1176.

Thomas, Sue. 1991. "The Impact of Women on State Legislative Policies." Journal of Politics 53:958-976.

Thomas, Sue. 1994. How Women Legislate. New York: Oxford University Press.

Times-Mirror Center for People and the Press. 1993. "The Vocal Minority in American Politics." Washington, D. C., July.

Verba, Sidney, Kay Lehman Schlozman, and Henry E. Brady. 1995. Voice and Equality: Civic Volunteerism in American Politics. Cambridge, MA: Harvard University Press.

Welch, Susan. 1977. "Women as Political Animals? A Test of Some Explanations for Male-Female Political Participation Differences." American Journal of Political Science 21:711-730.

Wilcox, Clyde. 1994. "The Year of the Woman." In Elizabeth A. Cook, Sue Thomas, and Clyde Wilcox, eds., The Year of the Woman: Myths & Realities. Boulder: Westview.

Williams, Leonard. 1994. "Political Advertising in the Year of the Woman: Did X Mark the Spot?" In Elizabeth A. Cook, Sue Thomas, and Clyde Wilcox, eds., The Year of the Woman: Myths & Realities. Boulder: Westview.

Wyatt, Justin. 1994. High Concept: Movies and Marketing in Hollywood. Austin: University of Texas Press.

Young, Iris M. 1994. "Gender as Seriality: Thinking About Women as a Social Collective." Signs 39:713-738.