The gender gap concept has been used to point out differences between men and women regarding attitudes about and participation in politics, policy views, and beliefs about the role of women in social and political life. Early studies identified a “traditional gender gap” in which women were more conservative than men, had less interest in politics, and consequently participated less (Almond and Verba Reference Almond and Verba1963). Beginning in the 1970s, the theory that women were politically conservative as a fixed, structural characteristic was challenged. New research argued that women developed new interests and began to vote more to the left as female labor force participation increased, resulting in a “modern gender gap” (Manza Reference Manza1998). This new gender gap has been attributed to a variety of factors including increasing modernization, which is particularly evident in postindustrial societies (Inglehart and Norris Reference Inglehart and Norris2003).

In regions like Latin America that have lower levels of economic development, education, female labor force participation, and secularism than do postindustrial societies, however, traditional gender differences may still prevail (Inglehart and Norris Reference Inglehart and Norris2003; Jennings Reference Jennings1998; Macaulay Reference Macaulay2006, 104–7; Rodríguez Reference Rodríguez2003). At the same time, structural changes in developing countries, as well as democratization and advances by women's movements, may promote movement toward a modern gender gap. Across the region, the mobilization of women was instrumental in pressuring for democratic transitions (Baldez Reference Baldez2002; Jaquette Reference Jaquette and ed.1991). With varying degrees of success, women fostered greater equality by pressuring for the establishment of state women's agencies, the passage of laws that criminalized domestic violence and addressed discrimination in the workplace, and the promotion of women's political representation through measures like gender quotas (Baldez Reference Baldez2003; Craske and Molyneux Reference Craske and Molyneux2002; Waylen Reference Waylen2000). While a considerable body of research has developed concerning the influence of the women's movement in pressuring for regime transitions, and later in achieving legal changes and political representation, the extent to which these demands and reforms have extended to the mass public is largely unexplored.

To understand how the mass public views gender issues and how these views are related to elite cues during the formative years of an emerging democracy, we explore the evolution of gender differences and assess women's and men's responses to elite cues regarding gender in the Dominican Republic. We examine differences in Dominican women's and men's attitudes and behavior from 1994 to 2004, an important decade in the country's democratic development, as well as a period of significant progress on gender-related policies. Specifically, we analyze the nature and evolution of gender differences in three important areas for women's participation: political interest, civic engagement, and support for women in politics. Political interest and civic engagement have often been positively linked to political participation, and differences in activism frequently translate into inequities in representation (Brady, Verba, and Schlozman Reference Brady, Verba and Schlozman1995). Gender gaps in interest and engagement have serious implications for equity in terms of participation inputs and policy outputs. Understanding gender disparities in these areas tells us if women are less engaged than men and why, and also illuminates reasons for inequities in representation.Footnote ¹ Moreover, interest and engagement are areas where traditional gender gaps have been observed and analyzed in other countries (Inglehart and Norris Reference Inglehart and Norris2003; Verba, Burns, and Schlozman Reference Verba, Burns and Schlozman1997).

Analyzing attitudes about women's political participation provides even clearer insight into gender dynamics. By assessing factors that shape support for women in politics, we directly examine the causal processes that lead to feminist political positions and analyze the depth with which Dominicans embrace such views. We also evaluate whether processes that produce support for women in politics are distinct among women and men. Together, these analyses of interest, participation, and feminist attitudes about politics enable us to understand the extent to which women and men engage the political realm at different levels, what factors most influence how these differences are produced in a developing democracy, and where gains may potentially be made in terms of equity in both participation and representation.

The Dominican Republic is an excellent context in which to examine these issues as sociopolitical trends of democratization and progress in gender equity mirrored those of other countries in Latin America. Similar to other Third Wave democracies, the 1990s were important in the Dominican Republic's democratic development. The first six decades of the twentieth century were marked by instability, a U.S. occupation (1916–24), and the brutal dictatorship of Rafael Trujillo (1930–61). A key figure of the Trujillo era, Joaquín Balaguer, then governed for 22 of the next 30 years, combining political stagnation with socioeconomic transformation. Our analysis begins at the end of Balaguer's rule, a period of extensive mobilization around the 1994 election, and extends into the current post-Balaguer democracy. Existing research suggests that political changes that accompany democratization are likely to facilitate support for, and the engagement of, women. Therefore, the Dominican Republic, as other countries in the region, would be expected to experience changes toward modern gender gaps over the recent past.

Dominican women attained significant political advances during this period. In the 1990s, political elites, led by coalitions of female civic leaders and legislators, encouraged women's rights and access to elected office (Gómez Carrasco Reference Carrasco and Julia2005). Motivated particularly by the Centro de Investigación para la Acción Femenina, a prominent women's nongovernmental organization, political parties incorporated women's issues to varying degrees for the first time during the 1990 electoral campaign (Cordero Reference Cordero1991). During the 1994 campaign, a broader coalition of women drawn from political parties, NGOs, and community organizations generated a more extensive plan to provide women with equal opportunities (Gómez Carrasco Reference Carrasco and Julia2005). By the second half of the 1990s, national-level politicians were using gender-inclusive language in their campaign speeches.

Important legal and bureaucratic changes favoring gender equity were also achieved, if not always fully realized. Funding for the Office for the Promotion of Women (Dirección General de la Promoción de la Mujer—DGPM), which had been established in 1982 within the Ministry of the Presidency, increased during the 1990s. Then in response to a long-standing demand of feminist groups, the DGPM was elevated to the cabinet level, becoming the Ministry of Women (Secretaría de Estado de la Mujer) in 1999 (CEDAW 1998).

In 1997, Congress passed two important laws promoting women's private and public rights. The first law (Ley contra la Violencia Intrafamiliar) established protections from domestic violence in accordance with the Inter-American Convention on the Prevention, Punishment and Eradication of Violence against Women, which the Dominican Republic ratified in 1995 (CEDAW 1998; Perez Reference Perez2005). The second law established a quota system requiring women to be 25% of the candidates in city council and Chamber of Deputies elections. In 2000, new laws raised the quota to 33%, established that male and female candidates should alternate in placement on party lists, and required parties to nominate a woman for either mayor or vice mayor.Footnote ² These measures reflected the influence of similar ones in other countries in the region (Htun and Jones Reference Htun, Jones, Craske and Molyneux2002; Krook Reference Krook2006). As a result of these reforms, the Dominican Republic now compares favorably to other countries regarding women in local-level offices and the legislature, though not in national executive-level cabinet positions (Valdés, Muñoz, and Donoso Reference Valdés, Teresa and Alicia2005).

A symbolic capstone of progress for women in the Dominican Republic was attained in 2000, when Milagros Ortiz Bosch was elected as the country's first female vice president. Overall during the 1990s and early 2000s, the discourse and actions of partisan elites were more favorable to women's participation than in prior periods. Yet progress did not continue in a linear direction. As we discuss more fully below, political elite discourse in fact shifted dramatically as the 2004 presidential campaign approached, becoming more confrontational and less positive toward women. This reversal allows us to test the susceptibility of mass attitudes about women in politics to shifts in elite cues, thereby exploring the extent to which acceptance or rejection of women's rights and representation among elites is paralleled among the public. Our goal in this article, beyond confirming broad expectations about shifts in gender gaps, is to analyze how mass attitudes evolve and to assess sensitivity to changes in elite cues. These issues remain relatively understudied in emerging democracies and have important analytical and policy implications.

One of our most important findings is that the process of attitudinal change is highly uneven, a factor sometimes minimized in modernization type explanations (Adams and Orloff Reference Adams and Orloff2005; cf. response by Inglehart and Norris Reference Inglehart and Norris2005). Prior to 2004, positive cues by elites enhanced support for women in politics among party followers, but in 2004, overall levels of support for women declined noticeably and negative cues by party leaders hurt perceptions of women, though only among male followers. This is congruent with other research regarding the importance of contextual cues impacting political attitudes and behavior (Atkeson Reference Atkeson2003; Sapiro and Conover Reference Sapiro and Conover1997), and suggests that advances in men's support of gender issues are less stable than those among women. We discover that education and age influence women and men differently and in ways not previously highlighted in the literature. While there are no age differences among women in their attitudes about women in politics, men's support for women's political involvement increases for men through middle age. This result does not support a simple image of generational replacement as the major explanatory factor driving attitudinal change. Our analysis emphasizes the importance of elite issue leadership and education as central to increasing support for women in politics.

THE GENDER GAP IN THE DOMINICAN REPUBLIC

To analyze the evolution of gender gaps in Dominican democracy, we employ data taken from four national public opinion surveys—the DEMOS surveys—conducted in 1994, 1997, 2001, and 2004.Footnote ³ These landmark surveys asked questions about civic and political engagement, views about women, and social and political values for which data had not previously been available. The surveys were funded by the U.S. Agency for International Development's Democratic Initiatives Project, administered by the Pontificia Universidad Católica Madre y Maestra in Santo Domingo. To draw a sample representative of all Dominican citizens age 18 or over, a multistage, stratified sampling method was used.Footnote ⁴ Interviews were conducted in respondents' homes on both weekdays and weekends, guaranteeing that the samples included people of different backgrounds and socioeconomic levels, and care was taken to make the survey instrument readily accessible. These efforts captured samples reflecting the Dominican population and ensured low nonresponse and refusal rates.Footnote ⁵

Table 1 compares the means for men and women on three additive scalesFootnote ⁶ that measure political interest, civic engagement, and support for women in politics, allowing us to assess the presence or absence of gender gaps.Footnote ⁷ The interest scale is generated from three items, which each range from 0 (none) to 2 (a lot) and measure interest in politics, frequency of political news exposure, and frequency of political conversations. The engagement scale tallies the number of organizations to which a respondent belongs.Footnote ⁸ Respondents earn one or two more points on the scale for sometimes or frequently attending neighborhood meetings.Footnote ⁹ Five items measure support for women in politics: Women should participate more in politics, women candidates inspire as much confidence as men, women have as much governing capacity as men, women should participate in politics the same as men, and “politics is for men.” Agreement with the first four items earns one point on the scale; disagreement with the final item earns one point.Footnote ¹⁰

Table 1 Gender gaps in interest, engagement and support for women in politics, 1994–2004

Note: Reported means are weighted to correct for survey sampling effects.

^a Positive scores indicate traditional gender gap; negative scores indicate modern gender gap.

Standard errors are available from authors.

*p < .05; ** p < .01 indicate mean for male respondents is significantly different from mean for females using two-tailed t-tests.

Political interest increased slightly between 1994 and 2004. Interest among men remained fairly constant, but women's interest grew by 20%. Although a traditional gender gap in political interest persists in the Dominican Republic, the gap narrowed over this time period. Examining civic engagement, men participated significantly more than women in all four surveys, although the gap narrows between 1994 and 2004. Overall levels of civic engagement declined by close to 50% over 10 years.Footnote ¹¹

We find limited evidence of gender gaps in support for women in politics during the 1990s. But in 2001, a modern gender gap emerges as support for women's participation grows more rapidly among women than men. On the scale and on each item from which it is constructed, women are more egalitarian than men in the twenty-first century. For both genders, egalitarianism increases over the decade. But within this general pattern an important qualification is necessary because in 2004, support for women in politics declines slightly, particularly among men. The greater reversal in support for women in politics among males contributes to the significant modern gender gap in 2004. Following other research, we believe this stems from contextual factors that can have important impacts on attitudes about gender (Atkeson Reference Atkeson2003; Sapiro and Conover Reference Rodríguez1997). In 2004, the country's first female vice president governed and remained closely identified with a highly discredited administration, and she participated in an electoral campaign with more explicit machista overtones than previous ones. We assess the effects of elite cues in the multivariate analysis that follows. The changes in attitudes about women in politics suggest a shift toward egalitarianism, but the step back in 2004, especially by men, signals that these advances are not well established.

WOMEN AND POLITICS IN THE DOMINICAN REPUBLIC

Beyond describing the nature and evolution of gender gaps, we also seek to explain the causal factors that have shaped support for women in politics and the engagement of women versus men. We employ multivariate regression to analyze political interest, civic engagement and support for women's participation in politics, using the scales we introduced previously.Footnote ¹² We begin by examining political interest and civic engagement to determine the factors that encourage people to opt into involvement and to determine whether the gender gaps in political interest and civic engagement observed in the bivariate analysis persist when controls are introduced.

In addition to gender, we consider three sets of independent variables that we expect to influence political interest and civic engagement: motivations, integration into public life, and sociodemographic variables. The first set of variables measures motivations to be interested in politics or to participate in civic organizations. We hypothesize that economic downturns or failure to deliver basic services may provoke engagement in order to evoke change, while contentment with government performance will lead to less engagement (Finkel and Opp Reference Finkel and Karl-Dieter1991). To assess the role of dissatisfaction with government in motivating engagement, we include two independent variables: service delivery evaluations and economic assessments. We measure service evaluation with an eight-point scale that combines assessments of seven government services and views of the economy with an item that asks respondents to evaluate their family's economic status on a five-point scale ranging from very bad to very good.

Second, we hypothesize that integration into public life will produce more engagement. We use employment status to measure integration—people who are working will have more opportunity and impetus to be engaged than will the unemployed. Those who are not working are less likely to connect with other individuals or groups that promote participation (McDonough, Shin, and Moisés Reference McDonough, Shin and Moisés1998).Footnote ¹³ Interest in the public realm is also likely to translate into active participation through civic engagement. Previous research has also demonstrated that both news exposure (Brehm and Rahn Reference Brehm and Rahn1997) and overall political interest (McDonough, Shin, and Alvaro Moisés Reference McDonough, Shin and Moisés1998) promote participation in neighborhood groups and voluntary associations. So we expect a general interest in politics, which captures news exposure as well as self-reported political interest, to encourage greater civic participation. Therefore, political interest is included as an independent variable in the civic engagement analysis.Footnote ¹⁴

Sociodemographic variables are the third set of factors associated with political and civic engagement. We expect that as people become older (Strate et al. Reference Strate, Parrish, Elder and Ford1989),Footnote ¹⁵ more educated (Brehm and Rahn Reference Brehm and Rahn1997), and accumulate more resources (Brady, Verba, and Schlozman Reference Brady, Verba and Schlozman1995; Brehm and Rahn Reference Brehm and Rahn1997), they will be more engaged. We also take account of the possibility that these factors may operate differently among men than they do among women, as other research has shown (Howell and Day Reference Howell and Day2000; Jennings Reference Jennings1998; see Bolzendahl and Myers Reference Bolzendahl and Myers2004 for a partial dissent). We consider interactions between gender and three socio-demographic factors: education, age, and socioeconomic status. The effects of marital statusFootnote ¹⁶ and place of residence are also tested (Brehm and Rahn Reference Brehm and Rahn1997).

POLITICAL INTEREST AMONG WOMEN AND MEN

Table 2 presents the multivariate analysis of political interest. The findings indicate that even with the inclusion of other independent variables, gender maintains a statistically significant relationship with political interest in all four years. This finding is consistent with other research that has found that although women now tend to vote as often as men, they still have less interest in politics (Inglehart and Norris Reference Inglehart and Norris2003; Verba, Burns, and Schlozman Reference Verba, Burns and Schlozman1997). The difference between women and men does decline significantly over the period, suggesting that some advances in eliminating the traditional gender gap are being made.

Table 2 OLS regression: Political interest in the Dominican Republic, 1994–2004

Note: Ordered logit yields the same substantive results. ^aReference category: 40–49 years.

** p < .01; * p < .05. Significance tests are two-tailed. Standard errors in parentheses.

The analysis considers the mobilizing potential of frustration with government performance and finds that negative evaluations of government services are associated with greater political interest in 1994 and 2001. But there is no effect for service evaluations in 1997 or 2004, and economic evaluations have no significant impact in any year. There is no support for the idea that people integrated in the public sphere through employment will be prompted to take an interest in politics. In terms of structural factors in 1994 and 1997, socioeconomic status has a positive effect that translates into a difference of about .75 points on the political interest scale between the poorest respondents and the most prosperous ones. But this effect dissipates in 2001 and 2004. Marital status and place of residence have no significant influence.

We include interactions for both age and education with gender in order to explore whether these factors have different effects among men and women.Footnote ¹⁷ Following conventions for interpreting the substantive meaning of interaction terms, Table 3 displays the calculation of the marginal effects (Brambor, Clark, and Golder Reference Brambor, Clark and Golder2006) of age and education among men and women, as well as the conditional standard errors (Friedrich Reference Friedrich1982).Footnote ¹⁸ These calculations enable us to determine how these factors operate among men and women and to assess whether there are gender differences in their effects on political interest. We measure age using five groups to allow for the possibility of a nonlinear effect for age, with respondents in their 40s as the reference group. The marginal effects of age on political interest presented in Table 3 demonstrate that men under 40 are significantly less interested in politics than men in their 40s, but that there are no significant differences in interest among men over 40. We also find no differences in political interest among women of different ages, except in 1997 when women under 30 are significantly less interested than are women in their 40s. So men's political interest increases until they reach middle age, when it levels out, while women's interest tends not to vary with age.

Table 3 Marginal effects of age and education on political interest

Note: Calculations based upon centered variables. Conditional standard errors are in parentheses. ** p < .01; * p < .05.

^a Reference category: 40–49.

As hypothesized, education has a strong positive influence on political interest. Table 3 also reveals that the effect of education in promoting interest is greater among men. The difference is present in all four surveys, but is most notable in 2004, when the coefficient for men is .18 points higher than for women. This sizable difference in the effect of education suggests that while education promotes political engagement overall, there may be something about the educational process in the Dominican Republic that encourages men to take more interest in politics than women, even as men also typically increase their political interest from their youth into their 40s, while women's interest remains more fixed.

CIVIC ENGAGEMENT AMONG WOMEN AND MEN

The analysis of civic engagement presented in Table 4 indicates that the factors contributing to civic participation in the Dominican Republic are distinct from those that shape political interest.Footnote ¹⁹ Unlike interest, gender has a more modest impact on civic engagement, with small effects in the first three surveys and no significant effect in 2004.Footnote ²⁰ Also, there are no gender differences in the effects of the independent variables on engagement as there are for interest. Interacting gender with education as well as with other variables not shown in the final analysis here reveals no significant gender differences in causal processes.

Table 4 OLS regression: Civic engagement in the Dominican Republic, 1994–2004

Note: A logged version of the scale is used as the dependent variable in this analysis to avert problems with heteroskedasticity. Logged version range:

0 to 2.4. Tobit Analysis produces the same substantive results.

^a Reference category: 40–49 years.

** p < .01; * p < .05. Standard errors in parentheses. Significance tests are two-tailed.

We hypothesized that dissatisfaction with government performance would motivate greater participation but find only limited support for this idea. Pocketbook evaluations have a negative relationship with engagement in 1997, but in 2001 the relationship is positive, and in 1994 and 2004 there is no effect. Additionally, service evaluations actually have a statistically significant, positive influence on engagement except in 1997. On the other hand, the analysis supports our expectations concerning integration into public life. Both the employed and the politically interested are inclined to be active in civic organizations. These variables have some of the most consistently significant and substantively important effects here. Except for 1994, the employed, on average, participate in one more civic group than do the unemployed. Those at the top of the political interest scale are in two more civic groups than those with no political interest.

There are significant effects for age and education. Following previous research (Strate et al. Reference Strate, Parrish, Elder and Ford1989), the youngest Dominicans are the least engaged, and as they reach old age civic engagement declines. Our analysis also coincides with previous research in the finding that more education (Brehm and Rahn Reference Brehm and Rahn1997) produces more participation. Residence in rural communities also promotes greater levels of civic engagement.

To explore why the gender gap diminishes significantly in the multivariate analysis of civic engagement but not in the analysis of political interest, we conducted additional analyses not shown here. Without political interest as an independent variable in the civic engagement regressions, the gender gap in engagement increases considerably. The findings concerning the other independent variables do not change when political interest is dropped, but gender becomes significant in all four surveys, with women participating in about one civic organization less than men. But when controlling for political interest, the gender gap in engagement practically vanishes.Footnote ²¹ Politically interested women are not less likely to participate in the public realm than similarly interested men, and so enhancing women's interest in politics would serve the dual goal of reducing gender differences in interest and in civic engagement. As greater gender parity in political interest is achieved, we are likely to observe increased equality in civic activism.

EXPLAINING SUPPORT FOR WOMEN IN POLITICS

In the final portion of the analysis, we examine the scale measuring attitudes about women's political participation, which includes both general questions about women's political participation and more specific questions about the suitability of women as candidates and in public office. We consider three sets of hypotheses. First, we expect that views of women's roles in society more broadly will shape perceptions of the political involvement of women (Banaszak and Plutzer Reference Banaszak and Plutzer1993). Politics is often viewed as a complicated and competitive domain, which should be reserved for men. But people who hold more generally egalitarian views of women will more likely be open to women's equal participation in politics. If respondents believe that women should take active and influential positions in the economy and the family, then they are more likely to support women's active participation in politics as well. To assess how views about women in society shape support for women in politics, we use two variables: support for women working and support for women's participation in household decisions. Support for women in the workplace is measured using a question that asks whether women should only work if men's income is not sufficient. Respondents who disagree with limiting women's employment in this way score a one on this variable. Support for women in the household is assessed with a question that asks respondents who should make household decisions: men, women, or both.

The second set of hypotheses concerns elite cues. Political leaders are likely to project an array of signals concerning how society should be organized and how politics should work. If political leaders take specific, well-publicized steps to promote women's participation in politics or use more egalitarian discourse, these practical and symbolic overtures of inclusiveness may promote broader support for women in politics either as ordinary citizens or as candidates and elected officials. Alternatively, the exclusion of women from important political posts, the marginalization of women in such posts, or the use of machista rhetoric may encourage more sexist attitudes among the general public (Gillespie and Spohn Reference Gillespie and Spohn1987). Elite cues are most likely to shape the attitudes of people such as activists or party members, who are especially connected with and attuned to the political realm. For this reason, previous studies include measures of partisan identification as a means of assessing the extent to which a respondent's partisan ties might influence his or her views about women in politics (Banaszak and Plutzer Reference Banaszak and Plutzer1993; Gelb Reference Gelb1989).

To test for elite influence, we incorporate measures of political connectedness into our analysis. We employ partisan affiliation as an independent variable to assess how respondents' ties to politics, and specifically their affinity for political elites, shape their views of women in politics. We expect people with stronger partisan ties to be more influenced by either substantive or symbolic changes in elite positions on matters pertaining to women's roles in politics. As discussed in the introduction, the period prior to 2004 was one in which Dominican political elites, led by coalitions of female civic leaders and then legislators, encouraged women's rights and women's access to elected office. Significant legislation to combat domestic violence was approved, state policy machineries to promote women's interests were introduced and strengthened, and gender quotas for local and legislative elections were passed.

We anticipate a change in the nature of elite influence about women's issues in 2004 relative to previous years, however.Footnote ²² In the 2000 presidential election, Hipólito Mejía was the presidential candidate of the major opposition Dominican Revolutionary Party (PRD). He was known for his strong, even vulgar, discourse in contrast to the country's other national-level male politicians. But during the 2000 campaign, his discourse was tempered and its impact mitigated by the fact that he enjoyed a comfortable advantage in the polls and that Milagros Ortiz Bosch was his running mate. Ortiz Bosch was a prominent politician within the PRD who had been the only woman elected senator for the 1994–98 period, and she had played an important role in passing legislation advancing women's interests.

In contrast, throughout the 2004 campaign, Mejía's aggressive and machista rhetoric surged, with the leading opposition candidate occasionally responding in kind. In 2002, the PRD-dominated Congress approved a constitutional amendment permitting immediate presidential reelection, even though the party had always opposed reelection. Over 2003 and into early 2004, Mejía became engaged in a bitter intraparty conflict regarding the selection of the party's nominee for the 2004 election, which was only for the presidency and thus focused on the personalities of the leading candidates.Footnote ²³ Ortiz Bosch had strongly opposed Mejía's reelection and sought the party's nomination herself. But in the end, she endorsed and campaigned for Mejía, a candidate who was heading an unpopular administration in the midst of a severe economic crisis; also under Mejía, progress on various gender equity goals stalled despite her position in the administration.Footnote ²⁴ Ortiz Bosch's ties to this unpopular administration and its machista rhetoric only served to undermine the credibility of the most prominent female politician in the country. In early 2004, Mejía turned from the intraparty conflict to a campaign against his opponent, Leonel Fernández. This campaign was distinctly harsher, more sexist, and more confrontational than in 2000, as polls indicated Fernández enjoyed a wide advantage over Mejía (Fernández ultimately won by a margin of over 25%).

The 2004 campaign represents a shift from previous campaign discourse, which is why we argue that cues from party leaders undermined advances in attitudes toward women in politics made throughout the 1990s.Footnote ²⁵ As a result of this context during the 2004 survey, we expect respondents with party ties to be less inclined to support women's political participation than people without partisan affiliations.

Our expectations are in line with other research that has found contextual cues to have a strong influence on attitudes and behaviors pertaining to women's political engagement. Atkeson (Reference Atkeson2003) finds that the presence and visibility of women as politicians serve to legitimize women's participation in politics more generally, and Sapiro and Conover (Reference Conover and Gray1997) find that particular electoral cues and messages impact the gender basis of electoral behavior. Therefore, we hypothesize that partisan affiliation is likely to promote support for women initially but to hurt perceptions of women in politics by 2004. We separate party members from partisan sympathizers in the analysis in order to assess the influence of different attachment levels.

The third group of potentially significant factors in shaping support for women in politics involves life experiences. We account for sociodemographic characteristics that may affect views of women's role in politics. We include gender as a predictor and expect that men will be less supportive of women's participation, particularly as the modern gender gap emerges (Gillespie and Spohn Reference Gillespie and Spohn1987). We examine life-cycle effects by exploring differences among age groups (Davis and Robinson Reference Davis and Robinson1991; Welch Reference Welch1975). Education is incorporated, with the expectation that more education promotes more openness to women's participation (Banaszak and Plutzer Reference Banaszak and Plutzer1993). Finally, we consider marital status and income as predictors of feminist attitudes (Baxter and Kane Reference Baxter and Kane1995; Conover and Gray Reference Conover and Gray1983).

Previous empirical research (i.e., Banaszak and Plutzer Reference Banaszak and Plutzer1993; Inglehart and Norris Reference Inglehart and Norris2003), as well as the theoretical argument that the process of acquiring feminist values is distinct among men and women (Klein Reference Klein1984), suggest that it is important to allow for the possibility that these factors may have differential effects on men and women. For instance, education may have a stronger effect among women because they may more readily accept the egalitarian ideas that education presumably promotes (Baxter and Kane Reference Baxter and Kane1995). Therefore, we also consider interaction terms between gender and our independent variables to determine whether these variables have different effects among men and women.Footnote ²⁶

ANALYZING SUPPORT FOR WOMEN IN POLITICS

Table 5 presents the analysis of support for women's political participation. In all four surveys, support for women working and for women's participation in household decision making significantly increase support for women in politics. Respondents who thought women should be able to work if they choose score almost one point higher on the support for women in politics scale than do those who thought women should only work if a man's income was insufficient. Thinking that women should be involved in household decisions, either as equal partners or as primary decision makers, is also associated with more egalitarian views of participation. Supporting joint decision-making results in nearly a full point increase on the dependent variable over those who believe that only men should make household decisions. General attitudes about women in the family and society shape more specific views about women in politics.

Table 5 OLS regression: Support for women in politics in Dominican Republic, 1994–2004

^a Reference category: “Men Decide.”

^b Reference category: No partisan sympathies.

^c Reference category: 40–49.

** p < .01; * p < .05. Standard errors in parentheses. Significance tests are two-tailed.

Elite cues also have an important influence on support for women's participation. To assess the impact of elite issue leadership, we include variables for party membership and partisan sympathies; those with no partisan affiliation are the reference category.Footnote ²⁷ We interact partisan ties with gender to assess if the effects of elite cues are different between women and men. Examining Table 5, we immediately see that membership in a political party has a significant effect on views about women's participation. What is more, the nature of this effect varies from survey to survey in a pattern reflective of the predictions developed earlier based on the changing nature of elite cues on the issue of women's roles.

The first row of Table 6 parses out the marginal effects of party membership among women and men. As expected in 1994, party membership produces greater support for women's political involvement among both female and male respondents. Female party members scored .35 points higher on the women in politics scale than did women who were not affiliated with a political party, and male party members scored .29 points higher on the scale than did nonaffiliates. In 1997, the effect of party membership is diminished and marginally significant, but is positive as expected. Again in 2001, we see a positive relationship between party membership and support for women in politics, but here the effect is only significant among men.

Table 6 Marginal effects of party ties, age, and education on supporting women in politics

Note: Conditional standard errors in parentheses.

^a Reference group: 40–49.

** p < .01; * p < .05. Calculations use centered variables.

Three years later, though, elite cues no longer promote women in politics, and we observe that partisan ties now undermine men's support for women's participation. In 2004, both party membership and partisan sympathies actually reduce men's support for women in politics, a result that reflects the expectations about the pattern of elite influence from 2000 to 2004 under the Mejía administration. Machista rhetoric and elites' deemphasizing of women's contributions undermined men's support for women in politics. It is also interesting to note that although men continue to be influenced by elites throughout the decade, female party members in both 2001 and 2004 are not influenced by elite cues. This finding may be the result of women's greater likelihood of forming their own views about women in politics, rather than listening to elites, especially when elites are undermining their rights. Or perhaps broader gender-related socioeconomic and cultural dynamics have taken root more extensively across women than across men. In both surveys, women are significantly more likely than men to favor women's political participation.Footnote ²⁸

The final variables in Table 5 are sociodemographic: gender, age, education, socioeconomic status, and marital status. In 1994 and 1997, female respondents are no more likely to support women's political involvement than are their male counterparts. However, in the following years, a modern gender gap emerges and persists in the multivariate analysis; the gap is marginally significant in 2001 and clearly significant in 2004. The last five rows of Table 6 present the marginal effects of age and education among men and women. Only in 1994 do we find significant age effects among women. Women in their 20s and 50s are significantly less supportive of women's political participation than women in their 40s, while women over 60 are significantly more supportive. But age has no significant effects among female respondents in any other year. Among men, the effects of age are much more consistent. In every year but 2004, men under 30 are much less supportive of women in politics than are middle-age men. So while female respondents tend to maintain the same level of support for women in politics as they age, men's attitudes become increasingly positive until they reach their 40s, when their views stabilize.Footnote ²⁹

Education has a significant and substantial positive influence on support for women in politics. As we can see in Table 6, the effect is significant for both genders, but is particularly strong for female respondents. In every survey, women score at least .03 points higher than men on the attitudes-toward-women scale for each additional level of educational attainment. In 1994 when the effect is especially strong, women with at least some college education score 3.4 on the scale, while women with no education score only 2.0; uneducated men score 2.4, and even highly educated men score only 3.0. This finding suggests that education is an excellent way to raise support for women's political participation within the entire population, but particularly among women.

We only find significant effects for socioeconomic and marital status in 1994, where socioeconomic status has a positive relationship with egalitarian views of political participation, and where being in a marital situation has a negative relationship. In the subsequent surveys, these effects are no longer significant when controlling for other factors.Footnote ³⁰ Education and socioeconomic status are correlated; if education is removed from the model, socioeconomic status has a positive and significant effect in all four years.

The analysis of support for women's participation in politics points to several important factors. Particularly influential are general attitudes about women's roles, with more egalitarianism leading to more support for women's political participation. Elite cues have a significant impact on people closely connected to the political system through party membership, and men in particular are vulnerable to changes in elite cues. So the views that political elites promote concerning women's participation in politics have a significant influence on citizens' support for gender egalitarianism in politics. Finally, education, especially among women, and age among men substantially increase support for women's political involvement.

SHAPING GENDER DYNAMICS IN EMERGING DEMOCRACIES

Our analysis indicates that in the Dominican Republic between 1994 and 2004, traditional gender gaps shrank but remained in place in the areas of civic engagement and political interest. In multivariate analysis, gender differences in engagement narrowed, but the traditional gap in interest remained. In light of persistent gender gaps in political interest in other countries, this finding is not surprising (though still disturbing). Our multivariate analysis of civic engagement, though, underscores an issue that has largely not been emphasized in previous research. Ongoing gender differences in civic engagement, which are present in the Dominican Republic as well as in many other countries, may be attributed to lower levels of political interest among women. It would be valuable to pursue research in other new democracies in order to discern if political interest advances among women would likewise promote parity in engagement within other contexts. At the same time that traditional gender gaps diminished but persisted in interest and engagement, we find that a modern gender gap emerged with regard to support for women in politics. This progress from traditional to modern gender gaps during a transformative period in an emerging democracy is in line with expectations generated by previous research. Progress toward a modern gender gap is more advanced in Dominicans' attitudes than in their actions.

From a comparative and policy perspective, one of the most interesting findings from the multivariate analyses concerns the differences in the effects of age and education among women and men. Consistently in all four surveys, men develop greater political interest as they age into their 40s, but women largely do not. We find similar results with regard to support for women in politics; controlling for other factors, men's attitudes become more positive as they age, whereas women's views are more similar across age groups. This finding contradicts some previous research, which has found that younger people hold more feminist views (Plutzer Reference Plutzer1991; Welch Reference Welch1975) and that generational replacement will reduce or eliminate traditional gender gaps (Inglehart and Norris Reference Inglehart and Norris2003). With consistent levels of political interest and feminist attitudes among women of all ages in the Dominican Republic, it is unlikely that generational replacement alone will lead to increases in interest or feminism among women over time. This divergence raises the possibility that generational replacement producing greater gender equity is not an inevitable process, but rather hinges on other factors that produce the generational difference among women that has been observed in some countries, but not here.

We also observed education's powerful effect on women's attitudes about their role in politics, far stronger than education's influence on men. But in the case of political interest, more education actually widens the traditional gender gap, as the effect of education in boosting interest is stronger among men. Education in the Dominican Republic appears to encourage feminist attitudes among women, without prompting them in an equivalent manner to become more engaged in politics.

Finally and perhaps most intriguingly, our analysis demonstrates that there are important gender differences in the effects of elite cues on support for women in politics. Specifically, we find that feminist attitudes among men align with shifts in elite behavior and discourse, while women's views are not as influenced by contextual factors. Visible elite support for women's political involvement in the 1990s promoted similar support among Dominicans who were most closely tied to these leaders—party members. But during the early 2000s, some political leaders stepped back from active promotion of women's participation in politics and embraced machista rhetoric, discouraging respect for women's political involvement. This served to undermine support for women in politics, particularly among male party members who may have been less firmly committed to the ideal of women's involvement and more susceptible to negative elite influence. So while positive elite cues prompted greater support for women in politics among both women and men, negative elite signals caused men to question the benefits of women's political participation. Together with previous research (Atkeson Reference Atkeson2003; Sapiro and Conover Reference Sapiro and Conover1997), this finding emphasizes the importance of elite influence in promoting or undermining gender egalitarianism and indicates that gender inequities are not entirely attributable to socialization processes. So even as women's movements and other civic groups continue to press for greater gender equity and a more active role for women in politics, these findings suggest how crucial it is that the rhetoric and actions of political leaders constantly enhance respect for, and the practice of, women's political rights.

Our findings, together with studies of women's rights in other emerging democracies (LaFont Reference LaFont2001; Rodriguez Reference Rodríguez2003), make clear that progress on women's issues is not unidirectional. It cannot be taken for granted as the ineluctable consequence of broader socio-economic processes or of democratization, but is susceptible to elite influence and therefore requires continued activism and political leadership.