A Different Hue of the Gender Gap: Latino Immigrants and Political Conservatism in the United States

Abstract

Using the 2012 Latino Immigrant National Election Study, we investigate gender differences in the liberal-conservative identification of Latino immigrants. We assess differences between Latino immigrant men and women in ideological ratings and consider two explanations for a different hue of the gender gap in political ideology. One emphasizes women’s greater social conservatism compared with men; the second considers whether and how gender differences in political ideology shift with longer U.S. residence. We find that Latinas are more politically conservative than Latinos, net of other factors, and that relationships between different social issue predictors, or length of U.S. residence, and liberal-conservative self-identification are gendered.

In recent history, gender gaps in U.S. politics have become commonplace. Although not viewed as a politically salient characteristic for much of the twentieth century, gender is now treated by political pundits, media analysts, and scholars alike as a key attribute that differentiates political attitudes and voting behavior. Generally speaking, compared with men, women are more likely to vote, support Democratic candidates, and hold different views on domestic and foreign issues. As a result, since the 1980s, many studies have examined the extent and determinants of the gender gap in the United States and other industrialized nations (see Verba, Burns, and Schlozman 1997; Kauf-mann and Petrocik 1999; Inglehart and Norris 2000; Arceneaux 2001; Kaufmann 2002, 2006; Paxton, Kunovich, and Hughes 2007; Clark and Clark 2008).

Although prior studies have suggested a number of explanations for these gender differences, large-scale and sustained growth in the immigrant population may complicate such explanations in the United States. Not only have immigrants increased in their share of the U.S. population since 1960, from approximately 5 to 13 percent in 2012, the foreign-born population has grown by 31 percent, from thirty-one to forty-one million, since 2000. Alongside this trend is growth in the Hispanic-Latino population. In 2010, 16 percent (50.5 million) of the U.S. population was of Hispanic or Latino origin, up from 13 percent (35.3 million) in 2000 (U.S. Census Bureau 2011). These changes have occurred during a period of gender-balanced U.S. immigration, although immigrant gender composition varies substantially by national origin (Donato et al. 2011; Donato and Gabaccia 2015).

In this paper, we focus on Latino U.S. immigrants and examine the relationship between gender and political ideology. Using the 2012 Latino Immigrant National Election Study (LINES), a new dataset about Latino immigrant political behavior and participation, we investigate gender differences in the liberal-conservative identification of Latino immigrants. We carefully assess differences between Latino immigrant men and women in ideological ratings and consider two possible explanations for a different hue of the gender gap in political ideology. One explanation emphasizes women’s greater moral conservatism. The second considers whether and how gender differences in political ideology shift with longer U.S. residence. As a whole, our analysis is part of a growing body of scholarship aimed at understanding how the Latino gender gap in U.S. politics varies across race, ethnicity, and nativity (Conway 2008; Bejarano, Manzano, and Montoya 2011; Bejarano 2013).

LITERATURE REVIEW

This section reviews studies on the gender gap in U.S. politics and focuses on those that examine the Latino gender gap and its determinants. We begin by presenting relevant findings from the Pew Hispanic’s 2011 National Survey of Latinos (NSL). Paul Taylor and his colleages (2012) report that Latinos nationwide have more liberal political views than the U.S. general population, but that Latinos hold more conservative beliefs on some social issues, such as abortion and homosexuality. Nativity further nuances these findings, with foreign-born Hispanics more likely than their U.S.-born counterparts to describe their views as conservative (35 versus 28 percent, respectively), and with U.S.-born Hispanics more likely than the foreign born to report being liberal or very liberal (34 versus 27 percent, respectively). Although this report published no gender differences, other work describes Latina immigrant women as being more politically conservative than their male counterparts (Bejarano 2013). As we see in the following section, this is a different hue of the gender gap and contrasts with the modern gender gap that Cal Clark and Janet Clark (2008) and others describe.

Gender Gap in Political Behavior and Ideology

Although gender was not viewed as central to understanding political behavior early in the twentieth century, political scientists began to reevaluate women’s role in politics in the 1980s (Tolleson-Rinehart 1992). A robust literature examines the gender gap in political beliefs, voting behavior, and partisanship toward different policy issues in the United States and elsewhere (see Verba, Burns, and Schlozman 1997; Kaufman and Petrocik 1999; Inglehart and Norris 2000; Arceneaux 2001; Kaufman 2002, 2006; Paxton, Kunovich, and Hughes 2007; Clark and Clark 2008). In the United States, a significant gender gap in voting first appeared in the 1980 presidential election, with more women than men favoring the Democratic presidential candidate Jimmy Carter (CAWP 2012). Since then, gender differences in voting and other political behavior have “become a permanent part of the American political landscape” (Clark and Clark 2008, 3). Compared with men, women are more likely to vote, support Democratic candidates, and hold liberal views on policy issues.

A number of factors explain the gender gap in U.S. politics. One is related to modernization. As national economies modernize and develop, women complete higher levels of education and increasingly participate in the formal sector of the economy. These changes accompany cultural shifts about gender, which are associated with a rise in feminism, and together they have political consequences that include, in some nations, a gender gap in political behavior and attitudes (Manza and Brooks 1998; Inglehart and Norris 2003; Paxton, Kunovich, and Hughes 2007). For example, in the United States, rising numbers of women in the workforce led to a “cumulative net shift among women” supporting Democratic presidential candidates (Manza and Brooks 1998, 1259).

A gender gap in politics may also arise from shifts in men’s behavior and attitudes rather than women’s. For example, if men are increasingly identifying as politically conservative or shifting from being Democrats to Independents, or if they are less likely to turn out and vote in elections, these shifts may help explain the gender gap. In fact, a recent Pew Research Center (2012) study shows that although women are more likely than men to identify as Democrats, this gender gap has not shifted since 1990. What has changed in recent years is men’s identification as Independents. Since 1990, men have increasingly identified as Independents and been more likely than women to do so.

Using data from the American National Election Studies (ANES), Barbara Norrander and Clyde Wilcox (2008) document that since 1972 men have become more conservative than women, and that among women ideological polarization is higher than in the past. Since the 1970s, ideological differences between working, middle-, and upper-class men have disappeared. However, during the same period, women with high levels of education have become more polarized from those with less education. Highly educated women are both more liberal and more numerous than their counterparts with less education. Interestingly, cultural conflicts between religious social conservatives and more educated secular liberals explain some of the ideological identity differences between women and men. Although women and men respond similarly to some cultural issues, women’s ideological identity was more likely than men’s to be based on attitudes about abortion, and men’s identity was more likely based on class concerns, though this latter finding holds less now than in the past.

Latino Gender Gap

Susan Welch and Lee Sigelman (1992) were the first to examine Latino gender differences in political attitudes, using national data to compare them with blacks and whites. The authors found that Latinas were more ideologically liberal and more supportive of the Democratic party than men, but that the Latino gender gap was smaller than that for blacks and whites. Not long after, Lisa Montoya (1996) expanded on their work by examining the gender gap among Latino immigrants from Mexico, Puerto Rico, and Cuba. Using data from the 1989 Latino National Political Survey, she found relatively small differences between men and women in political ideology, party identification, and presidential voting.

Recent work by Christina Bejarano and by Bejarano, Sylvia Manzano, and Celeste Montoya (2013, 2011) focuses on the Latino gender gap in U.S. politics and considers how the gender gap varies by race and ethnicity and by immigrant generational status. Both studies use data from the 2006 Latino National Survey to examine generational differences in Latino political ideology. Findings are that, with longer U.S. residence, a gender gap emerges among immigrants in their propensity to maintain close ties with their origins. This finding is consistent with studies by Sherri Grasmuck and Patricia Pessar (1991), Pierrette Hondagneu-Sotelo (1994), and Cecilia Menjívar (2000), who report that Dominican, Mexican, and Central American women spouses preferred to remain in the United States, but their male husbands preferred to return to origin communities. Also important for our purposes is that men’s and women’s responses about political ideology differ across generations. Relative to foreign-born Latinos—that is, the first generation—those in subsequent generations are more likely to identify as politically liberal. Moreover, with respect to a gender gap in political ideology, although immigrant Latinas rated themselves as more conservative than immigrant Latinos, by the fourth generation Latinas shifted to the left of men and reported their political ideology as liberal.

Explaining the Latino Immigrant Gender Gap

Only a few studies interrogate the Latino gender gap in political behavior and attitudes (Bejarano, Manzano, and Montoya 2011; Bejarano 2013). This work focuses largely on generational differences, asking how Latinos and Latinas born outside of the United States differ from the U.S. born. As a result, it tells us little about explanations for the nontraditional gender gap, whereby Latina immigrants are more politically conservative than men, and specifically whether and under what conditions this form of the gender gap shifts such that Latinas become more liberal than Latinos. Therefore, in line with Bejarano’s work, we begin by expecting that Latina immigrants will be more politically conservative than their Latino counterparts.

H₁: We expect that political ideology identification among Latino immigrants varies significantly by gender. Latina immigrants will be more politically conservative than men.

In our analysis here, we provide two possible explanations for this nontraditional gender gap in ideology among Latino immigrants. The first is about moral and social conservatism; it holds that women immigrants are more conservative than men on moral and social issues such as support for abortion or homosexuality. Presumably because of childhood socialization and adult experiences that are gendered along traditional and patriarchal lines, such differences in morally conservative views will help explain the gender gap in ideology. Therefore, we expect that effects for attitudes toward abortion, homosexuality, and other social issues depend on gender. Women’s conservative attitudes will increase women’s political conservatism, whereas men’s attitudes will not. This leads to two hypotheses:

H_2a: We expect that attitudes toward abortion or homosexuality will significantly predict political conservatism net of gender and other variables.

H_2b: We expect that effects for attitudes toward abortion or homosexuality will be conditioned by gender.

The second explanation captures political assimilation that occurs over time, and assesses whether and how Latino immigrants shift their ideological positions with more time spent in the United States. This theory holds that the gender gap in ideology is related to a carryover of attitudes from immigrants’ origin countries and length of U.S. residence. That political attitudes shift over time is an idea consistent with Zoltan Hajnal and Taeku Lee (2011), who document how partisanship among Latinos and Asian Americans changed with longer U.S. residence, but do not consider whether and how these assimilative effects are different for women than for men.

Grasmuck and Pessar (1991), Hondagneu-Sotelo (1994), and Menjívar (2000) describe strong gendered preferences among Latino immigrants to return to their countries of origin: with more time in the United States, Dominican, Mexican, and Central American women reported wanting to remain in the United States, whereas their husbands remained committed to eventually returning to their origin communities. Many of these women made economic contributions to their households, making traditional gender scripts more difficult to follow. In addition, compared with newly arrived Latina and Latino immigrants, Latinas with longer stays in the United States typically have greater access than men to schools and community organizations. These contacts may help socialize and teach immigrant women about political ideological identification and policy issues linked to being a Democrat versus a Republican, but in ways that are different from their male counterparts, whose understanding about what liberalism and conservatism mean may derive largely from connections to the workplace. Michael Jones-Correa (1998) shows that political socialization is gendered among Latino immigrants: the men in his sample oriented their politics toward their country of origin as women became involved in U.S. organizations. Thus, with more time in the United States, Latina immigrants may increasingly adopt U.S.-based political perspectives and, similar to U.S.-born women, become more liberal than men in ideological identification. Latino men, on the other hand, may become more conservative.

H_3a: We expect that length of U.S. residence will significantly predict political conservatism net of gender and other variables.

H_3b: We expect that effects for length of U.S. residence will be conditioned by gender.

Data and Methods

We examine whether and how gender differentiates Latino immigrants’ political ideology and explanations for this variation. To do so, we use data from the 2012 Latino Immigrant National Election Study, a nationally representative pre- and postelection telephone survey of immigrants from Spanish-speaking Latin American countries. Although similar to the 2012 ANES and using many comparable measures, the LINES sample is composed of 1,304 foreign-born Latino adults. The majority are not citizens; as a result, approximately 95 percent of the sample completed their interview in Spanish.

The preelection sample has 855 respondents, and the postelection wave has 886 respondents. Of those surveyed postelection, 435 respondents participated in the preelection survey and 451 respondents were new, surveyed only after the election. Because some key variables for this analysis are available only in the postelection data collection, we restrict our analytic sample to the 886 postwave respondents.

Following Rebekah Young and David Johnson (2010), we use multiple imputation procedures to recover missing data on the independent and dependent variables. To do so, we first created a series of replicate datasets that assigned imputed values for missing values. We separately analyzed each imputed dataset, and then pooled the replicate datasets to generate a single set of mean parameter estimates. We therefore used the multiple imputation by chained equation approach, to generate ten replicates using a series of univariate regressions. To maximize the size of our analytic sample, we imputed values for the following variables: age; naturalization; years of education; income; year of U.S. arrival; marital status; country of birth; feelings toward abortion, gays and lesbians, and the police; perceived discrimination variables; and liberal-conservative ranking.¹ With multiple imputation, we recovered relevant values for the complete postelection sample (N=886).²

Dependent Variable

Our dependent variable is liberal-conservative ideological ranking; it is measured by asking respondents how they would describe themselves. Respondents answered with one of the seven response categories: 1 = extremely liberal, 2 = liberal, 3 = slightly liberal, 4 = moderate, 5 = slightly conservative, 6 = conservative, and 7 = extremely conservative. If respondents did not answer this question the first time it was asked, they were then asked what they would consider themselves if they had to choose between liberal and conservative.³

Although prior studies suggest that liberal-conservative self-identification has been a poor predictor of party identification and policy stances (see Campbell et al. 1960), recent work suggests that partisanship among U.S. voters is now more strongly related to liberal-conservative identification (Lewis-Beck et al. 2008; Bafumi and Shapiro 2009). Because of rising polarization among U.S. voters, self-reported ideological orientations and political partisanship are more overlapping than in the past. Yet, among immigrants, little is known about whether liberal-conservative self-reporting correlates with party identification or policy stances. Therefore, to help justify our use of liberal-conservative self-identification as a dependent variable, we examined its relationship to difference in support for Obama and Romney as presidential candidates in 2012. Net of controls, we find a robust negative relationship between liberal-conservative identification and the difference in support; the more conservative Latino immigrant respondents are, the smaller the difference in support for the two presidential candidates. Table A1 presents coefficients from these models.⁴

Independent Variables

In addition to gender (coded 1 = women, 0 = men), we focus on a set of variables that assess social conservatism. The first is respondents’ feelings toward abortion. We recoded this into a series of dummy variables: always permitted = 1 if respondents reported that abortion should always be permitted and 0 = otherwise (reference category); permitted only after need = 1 if respondents felt that abortion should be permitted only after need was clearly established and 0 = otherwise; permitted if rape or incest = 1 if they believed that abortion should be permitted only in cases of rape, incest, or danger to the woman’s life and 0 = otherwise; and abortion should never be permitted = 1 and 0 = otherwise. The second set of variables are feeling thermometers that measure favorability toward gays and lesbians, people on welfare, and the police, using a scale from 0 to 100, 0 indicating least favorable and 100 most favorable. For each variable, we collapsed responses into three dummy variables: less favorable where 1 = less than 40 and 0 = other; middle-of-the-road where 1 = 40 to 59 and 0 = other; and more favorable where 1 = 60 up through 100 and 0 = other. In the regressions, more favorable is the reference category.

The final group of indicators of social conservatism capture perceived discrimination toward Hispanics, immigrants, and—more generally—five different groups. For perceived discrimination against Hispanics and immigrants, we recoded responses into three categories: a lot or great deal, moderate or a little, and none. To measure overall perceived discrimination, we created a summary index that ranges from 0 to 5 and summed respondents’ responses to questions about whether they perceive discrimination against Hispanics, immigrants, blacks, gays and lesbians, and women. We then recoded this summary index: if respondents perceived discrimination against four or five of these groups, we coded a dummy variable as 1 = a lot or great deal and 0 = other. If they perceived two or three, we coded a second dummy variable as 1 = moderate and 0 = other; if they perceived one or less, we coded a third dummy as 1 = little and 0 = other.

Finally, to assess an assimilation explanation for the gender gap in political ideology, we use a variable that captures years since arriving in the United States as an indicator of assimilation. This information is available in continuous years, and for the respondents in our sample, it ranges from less than one year up to seventy-four years. We recoded this information into a set of four dummy variables: zero to five years (the reference category); six to ten years; eleven to fifteen years; sixteen or more years.

Control Variables

We include a variety of control variables. Age is in continuous years, and marital status is a dummy variable where 1 = currently married, 0 = other. We recode education into a dummy variable where 1 = some college or greater and 0 = less than college. We enter income in its original ordinal form whereby 0 =< $20,000; 1 = $20 to $39,999; 2 = $40 to $59,999; and 3 = $60,000 or more. Being naturalized is also a dummy variable (1 = naturalized and 0 = not naturalized), and national origin is a set of dummy variables representing Mexican, Cuban, South American, Central American, and Dominican born.⁵

ANALYTIC STRATEGY

We begin by describing the extent to which gender differences in liberal-conservative ideological rankings exist within categories of key independent variables.⁶ We then present ordinary least squares (OLS) regression models that examine whether and how gender affects ideological rankings (model 1), and how the gender effect shifts after controlling for other variables (model 2). To determine whether gender effects differ for those with strong versus weak opinions about abortion; more versus less favorable feelings about gays and lesbians, people on welfare, and the police; a lot versus little perceived discrimination toward Hispanics, immigrants, and overall; and for those with shorter versus longer U.S. residence, we introduce interaction terms between gender and these items (model 3). The interactions indicate whether indicators of social conservatism and assimilation vary significantly by gender net of relevant explanatory variables. Effects for attitudes toward abortion, feelings toward different groups, and perceived discrimination varying by gender is support for the social conservatism hypothesis. If the effect for length of U.S. residence varies by gender, this is support for our assimilation hypothesis.

RESULTS

Table 1 displays how liberal-conservative self-identification rankings vary by key independent variables for the total sample, men, and women. The first row describes baseline gender differences in political ideological rankings, ranging from one through seven. Consistent with prior studies, women were significantly more conservative than men (4.4 versus 4.1).

With respect to feelings about abortion, the only significant gender difference was for respondents who believed that abortion should be permitted only for rape or incest. Among these respondents, women were significantly more conservative than men (4.7 versus 4.3). Women whose feelings toward gays and lesbians were less favorable were significantly more conservative than men (4.9 versus 4.6, respectively). Regarding feelings about people on welfare and the local police, women were also more conservative than men. Women who had more favorable and middle-of-the road feelings toward people on welfare were more conservative (4.4 versus 4.1 and 4.4. versus 4.0, respectively). Similarly, those who felt more favorably and those with middle-of-the-road feelings were also more conservative than men (4.5 versus 4.2 and 4.5 and 4.0, respectively).

Women were also more conservative among those who perceived moderate discrimination against Hispanics (4.7 versus 4.1), high or moderate levels of discrimination against immigrants (4.3 versus 4.1 and 4.6 versus 4.2, respectively), and a lot or a little discrimination overall (4.7 versus 4.1 and 4.7 versus 4.3, respectively). Finally, significant gender differences appear among those who have lived in the United States between eleven and fifteen years, and more than fifteen years. For both groups, women were more conservative (4.3 versus 3.8 and 4.5 versus 4.2, respectively).

Table 2 presents three OLS regression models that predict liberal-conservative ranking and include feelings toward abortion. Model 1 includes only the effect for gender, and its direction is consistent with the significant difference found in table 1. Women are significantly more conservative in political ideology than men. In addition, controlling for other variables, including those for feelings toward abortion, model 2 shows that, net of these effects, women are more conservative than men. Overall, we see that older respondents are more, and that those with at least some college are less, conservative. More income also increases conservatism rankings. Feelings toward abortion affect liberal-conservative self-identification in expected ways. Respondents with more restrictive views of abortion—that is, those who believe it should only be allowed for rape or incest or it should never be permitted—were significantly more politically conservative than those who had a more permissive stance.

Table 1.

Average Self-Reported Liberal-Conservative Ranking

Source: Authors’ compilation based on McCann and Jones-Correa 2012.

*p < .1; **p < .05; ***p < .01

Model 3 adds interactions between gender and feelings toward abortion. Results show that the interaction coefficients are not statistically

Table 2.

Regression Models Predicting Self-Reported Liberal-Conservative Ranking (Feelings Toward Abortion)

Source: Authors’ compilation based on McCann and Jones-Correa 2012.

*p < .1; **p < .05; ***p < .01

significant predictors of liberal-conservative ranking. However, in this model, the coefficients for feelings toward abortion represent the effects of these feelings among men. Therefore, male respondents with more restrictive views of abortion were significantly more conservative.

Table 3 presents two panels with selected coefficients from regression models predicting liberal-conservative ranking with six indicators of social conservatism. Panel A displays models that include feeling thermometers for gays and lesbians, people on welfare, and local police. Model 1 includes only the predictor for female and shows that women are more politically conservative than men. The two sets of models that follow show the effects for feelings toward one of the three groups, net of other factors, and effects for the interactions between gender and each of the feeling thermometer variables, respectively.

With respect to feelings about abortion, model 2a shows that net of these and other variables, the gender effect remains. In addition, respondents who had middle-of-the-road and less favorable feelings toward gays and lesbians were significantly more conservative. Model 2b adds interactions between gender and feelings toward gays and lesbians, and by doing so asks whether and how effects for support for gays and lesbians are conditioned by gender. Although we expected that Latina immigrants would be more conservative about supporting gays and lesbians than comparable Latinos, coefficients for the interaction terms are not significant. Instead, men with less favorable views of gays and lesbians are significantly more politically conservative (b = 0.735, p < 0.01). In addition, the effect for women with more favorable attitudes toward abortion, seen in the female coefficient (b = 0.352, p < 0.10), suggests that they are also more politically conservative. Thus, women are more conservative in political ideology than men, and some evidence indicates that effects for support of gays and lesbians on liberal-conservative self-identification varies somewhat by gender.

The next set of models includes feelings toward people on welfare. Once again, net of other factors, the effect for female remains, and those with less favorable views of people on welfare are more politically conservative. After adding interactions between gender and feelings toward people on welfare, we find that women with less favorable views of people on welfare are less (not more) conservative. Moreover, the significant female effect (b = 0.585, p < 0.10) suggests that women with more favorable feelings toward people on welfare self-report as more conservative. These findings contrast with the one finding for men: those with less favorable views of people on welfare are significantly more conservative.

The final set of models in panel A contains feelings toward local police. Model 4a shows that the gender effect remains, net of feelings toward local police, with women more conservative than men. Furthermore, similar to model 3b on the effects for women, model 4b suggests that liberal-conservative ranking is conditioned by gender and feelings toward local police. Women who hold less favorable views of local police are significantly less conservative (b = -0.702, p < 0.10), and those who hold more favorable views of the police are significantly more conservative. Therefore, findings from the three models that include feelings toward gays and lesbians, people on welfare, and local police suggest that women are sometimes more likely than men to be morally or socially conservative, but not in predictable ways.

Panel B shows coefficients from OLS regression models predicting liberal-conservative ranking, and includes variables for perceived discrimination against Hispanics, immigrants, and an overall summary measure of discrimination. Model 2a includes perceived discrimination against Hispanics and shows that women are more conservative than men, net of controls and key predictors. In addition, respondents who perceived some or little to no discrimination against Hispanics were significantly more conservative than those who perceived a great deal. In model 2b, we include interactions for gender and perceived discrimination against Hispanics but find that these are not statistically significant. Furthermore, we lose the effect for women who perceive a lot of discrimination (-0.105 is not significant). The model has only one gendered effect: it appears in the effect for none or little perceived

Table 3.

Selected Coefficients Regression Models Predicting Self-Reported Liberal-Conservative Ranking: Six Indicators of Social Conservatism

Source: Authors’ compilation based on McCann and Jones-Correa 2012.

Notes: Controls for age, marital status, level of education, income, citizenship, years in the United States, and whether respondent is Cuban, South American, Central American, or Dominican.

* p < .1; ** p < .05; *** p < .01

Table 4.

Selected Coefficients for Regression Models Predicting Liberal-Conservative Rank

Source: Authors’ compilation based on McCann and Jones-Correa 2012.

Notes: Controls for age, marital status, level of education, income, citizenship, years in the United States, and whether respondent is Cuban, South American, Central American, or Dominican.

*p < .1; **p < .05; ***p < .01

discrimination, which is really the effect for men. Here it is significant, but only at the .10 level (b = 0.703).

The second set of models in panel B includes perceived discrimination against immigrants. Model 3a shows, again, that women are more conservative than men, controlling for characteristics including perceived discrimination against immigrants. In contrast, respondents who perceived moderate levels of discrimination were significantly more conservative than those who perceived a lot of discrimination. However, perceiving little discrimination against immigrants was not a significant predictor of liberal-conservative ranking, and model 3b describes no significant effects for interactions between gender and perceived discrimination.

Using a summary index to measure overall discrimination, the final set of models reveal that respondents who perceived at least some or little to no discrimination were more politically conservative. When we add interactions for gender and perceived discrimination (model 4b), the only gender effect we see is that women who perceive moderate levels of discrimination are more conservative. Thus, these models suggest that gender differences in perceived discrimination against individual groups do not predict liberal-conservative ranking, the exception being for women who perceive moderate levels of overall discrimination.

Table 4 presents another set of regression models predicting liberal-conservative rankings. To assess whether variation in political ideology is related to a gendered process of assimilation, we include several dummy variables for years of U.S. residence. The gender coefficient in this model is significant with and without controls. Interestingly, model 2 shows no significant effects for length of U.S. residence. Yet after entering gender*duration interaction terms, we see that effects for length of U.S. residence vary significantly by gender. Women who have been residing in the United States for six to ten years are significantly less conservative (b = -1.213, p < 0.10). However, we also see a positive effect (p < 0.10) for being female, suggesting that newly arrived women (those with fewer than six years of U.S. residence) are more politically conservative (b = 0.917).

Figure 1.

Predicted Liberal-Conservative Rankings

Source: Authors’ compilation based on McCann and Jones-Correa 2012.

Note: Derived from table 4, model 3.

To better visualize these effects, we generated predicted ideological ratings in figure 1, which offers a visual representation of model 3 from table 4. Examining gender differences in the height of the bars within each duration category, we see that women who have been in the United States for longer are less conservative than when they first arrive, the largest drop corresponding to six to ten years of U.S. residence. In contrast, men with longer U.S. residence seem to become more conservative as the years go by, except the eleven to fifteen year span (where we see a slight dip). Our results, then, suggest a process of gendered political assimilation among Latino immigrants. Net of other factors, among recent arrivals, men are more liberal than women, but as the years living in the United States accumulate, this difference shifts and narrows.

DISCUSSION

In this paper, we find that Latino immigrants illustrate a different hue of the gender gap in political ideology: Latinas are more politically conservative than Latinos. This gender gap appears to be a consequence of both gendered differences in moral conservatism and gendered political assimilation. Thus, feelings toward abortion, gays and lesbians, people on welfare, and local police, as well as perceived discrimination, are related to liberal-conservative self-identification. Not only are immigrant women born in Latin American countries more politically conservative, this gap holds only for recently arrived immigrants. After living in the United States for at least five years, Latinas report themselves as more liberal, and after fifteen years, Latino men view themselves as more conservative. Thus, this analysis complicates our understanding about Latino political behavior and attitudes. It suggests that we have much more to learn before predicting where Latinos will ultimately fall on the political spectrum.

Although beyond the scope of this paper, this analysis raises a number of important questions for future research. First, to what extent do these findings hold for other race and ethnic groups in the United States? Second, if immigration is a rupture that substantially transforms the lives of both women and men and challenges traditional gender roles, then what other gendered political consequences might it have? Since the early 1990s and passage of restrictive immigration policies, antiterrorism legislation, and the welfare reform act, crossing the Mexico-U.S. border without authorization has become more complicated and costly (Donato and Armenta 2011). Migration from many countries in Latin America slowed after the economic downturn that began in 2006. Given current xenophobic sentiment and a strongly polarized U.S. political system, future Latino political attitudes and behavior may link to whether immigrant populations replenish themselves (Jiménez 2008). If so, it may be that the presence of large and established first- and second-generation Latino immigrant populations will tighten national-origin boundaries between Latino groups and have gendered consequences for political identity and behavior.

Our results also raise questions about the mechanisms that produce them. For example, why should the effects of length of residence depend on gender? Do women become more politically liberal over time because they primarily engage with schools, which are nonprofit institutions, while men become more conservative because they primarily engage with for-profit self-interested institutions? Although consistent with our results, we have no way to assess this because the data used in the analysis are cross-sectional and ill suited to address the task.

Finally, these findings suggest that future research must explicitly consider how liberalconservative self-identification is complex, nuanced, and gendered. Along these lines, it would be informative to consider whether and how results would differ if the question that solicits liberal-conservative self-reported ideology from Latino immigrants used izquierda and derecha (left or right) instead of liberal and conservador (liberal or conservative). Understanding the challenges of surveying Latinos in the United States is a large field of study, one that attempts to assess differences in the validity of survey questions translated across languages (Pew Research Center 2015). Although a priori there is no reason to think that use of the latter indicator explains why women’s self-identification differs from men, it is possible that immigrants may interpret conservative and liberal differently than natives.

Latino immigrant ideological self-identification is important because Latino immigrants and their offspring will represent a large segment of the future political landscape in the United States. Yet they are not a monolithic group. These immigrants hail from a diverse set of national origins (even if Mexicans currently outnumber all others), and they and their children represent different and, at times, overlapping interests. Exactly what these differences mean for the Latino electorate in the United States is not yet clear, but the results in this paper suggest they are likely to include salient gendered consequences.