Religion and Gender Ideologies among Working-Age U.S. Latinas/os

Numerous studies have documented religious variations in gender ideology in the United States. Despite growth, diversification, and religious ferment among Latinas/os, few have investigated this topic within the Latina/o population. Drawing on insights from gender theory and prior empirical research, we develop several hypotheses regarding the links between religious affiliation, belief, and practice and three distinct domains of traditionalist gender ideology (respective beliefs in female domesticity, gender essentialism, and patriarchy) among U.S. Latinas/os. These hypotheses are tested using data from the Hispanic oversample of the National Survey of Religion and Family Life (NSRFL), a nationwide probability sample of working-age adults (ages 18–59). The results underscore the complex associations between multiple dimensions of religious involvement and specific facets of gender ideology among Latinas/os. Several promising directions for future research on this understudied population are outlined, and study limitations are identified.


Introduction
Gender inequality and its ideological underpinnings remain among the most vexing issues of the contemporary era.Over the past four decades, numerous studies have examined patterns, correlates, and trends pertaining to gender ideology.Much of this literature has centered on three specific areas: (a) beliefs about the appropriate roles of women and men in the home and family and in the public sphere (e.g., workplace and politics); (b) beliefs about the origins of gender inequality (e.g., innate versus socially defined); and (c) beliefs about the desirability of various social actions and policies aimed at ameliorating gender inequality (Davis and Greenstein 2009;Kane 2000).
Despite the large and growing literature on these topics, researchers have noted the limited body of evidence on racial and ethnic differences in gender ideologies and especially the paucity of information on gender ideologies among Latina/o Americans (Kane 2000).Addressing this gap is particularly important considering the rapid growth of the Latina/o population such that Latinas/os have recently surpassed African Americans as the largest minority population in the United States.By some estimates, Latinas/os account for more than half of the nation's population growth between 2000 and 2010 (Saenz et al. 2007).The existing literature has focused on a number of potential sources of variation seen in gender ideologies among Latinas/os, including national origin, gender, assimilation, and socioeconomic status (Kane 2000).Viewed broadly, this body of research has yielded discrepant findings.However, the potential links between religion and gender ideology among Latinas/os have received minimal attention from investigators (for an exception, see (Hunt 2001)).
The lack of attention to linkages between religion and gender among Latinas/os is surprising for several reasons, all of which underscore the significance of our study.First, religion, particularly the traditional salience of Catholicism, among most Latina/o subgroups has been widely recognized in previous research (e.g., (Bartkowski et al. 2012;Espin 1994)).Historically, Latinas/os have been among the most devout Catholics in the world.Second, there is considerable religious ferment among Latinas/os in the contemporary United States and elsewhere in Latin America.Religious change among Latinas/os is especially evident in the current growth of conservative (i.e., fundamentalist, evangelical, and charismatic) variants of Protestantism (Diaz-Stevens and Stevens-Arroyo 1998;Steigenga and Cleary 2007).Third, the significance of this study is underscored by the fact that, in the general U.S. population, these religious tendencies have been associated with greater support for gender inequality and patriarchal norms and practices (Hoffmann and Bartkowski 2008;Sherkat 2000).
Our study addresses this notable gap in the research literature in several ways, focusing specifically on beliefs among U.S. Latinas/os regarding the appropriate roles of women and men.First, we briefly summarize the key findings of previous studies on Latina/o gender ideology.We then review patterns of continuity and change in contemporary Latina/o religion and outline a series of theoretical arguments linking religious affiliation, practice, and belief with gender attitudes.Hypotheses based on this discussion are then tested using data from the Latina/o subsample (n = 801) of the 2006 National Survey of Religion and Family Life (NSRFL), a nationwide probability sample of working-age adults with oversamples of African Americans and Latinas/os.The findings are discussed in terms of the literature on (a) religion, gender, and family and (b) Latina/o gender ideology in the United States.

Latina/o Gender Ideology: Integrating Current Research with Recent Theoretical Developments
The studies that have explored gender among Latinas/os have emphasized two dominant themes within Latina/o culture; familism and machismo (Kopinak 1995;Hamilton 2002;Gonzalez-Lopez 2004).Placing a premium on female subordination, these two themes serve as foundational components that help to shape Latina/o gender ideology and family life (Unger and Molina 1999).Familism is one essential feature very much intertwined with Latina/o culture.Despite the relatively low average socioeconomic status (SES) among Latinas/os, their rates of marriage and marital stability have traditionally been higher than those of other ethnic groups (Bean et al. 1996;Raley et al. 2004).Research shows that, within Latina/o culture, commitment is regarded as more important than individual autonomy in relationships, and this is especially true for Latina women (Oropesa 1996).Since familism is founded on the supremacy of the father figure and the self-sacrifice of the mother, it is an expectation that Latina women will put the needs of the family before their own personal interests.Therefore, Latina women are expected to accommodate the demands of family members while ignoring their own needs and not expressing discontent over such unequal arrangements (Diaz-Guerrero 1955).Aside from holding strong aspirations for marriage, Latina women often perceive their personal value and fulfillment to be derived through their familial contributions.For many Latinas, childbearing and self-sacrifice loom particularly large (Ellison et al. 2013;Unger and Molina 1999).Research suggests that this orientation stems from the way in which the 'machismo' ethos has been deeply woven into Latina/o culture, emphasizing the honor and leadership of Latino masculinity and thereby devaluing women relative to men.Machismo even influences and is reflected in cultural preferences for sons over daughters (Gonzalez-Lopez 2004;Unger and Molina 1999).The Latina/o gendered double standard begins even before adulthood, as is evident in a key cultural rite of passage, the 'quinceanera'.This birthday ceremony is a process of transition from girlhood to womanhood, marking young girls as ready for marriage.One expectation often associated with this ceremony is that of 'sexual purity', that is, the young woman's virginity.No such rites of passage or cultural expectations are aimed at Latino boys of this age.
Consistent with these assumptions, the small existing body of survey-based evidence has found that, as a group, Latinas/os are less egalitarian in their attitudes toward gender when compared to non-Hispanic whites or African Americans (Harris and Firestone 1998;Strong et al. 1994;Wilkie 1993).Although some observers have also reported significant attitudinal differences toward gender exhibited by Latina/o national origin groups (e.g., (Montoya 1996)), others have found few meaningful differences of this sort (Harris and Firestone 1998).
Prior research has linked a range of factors with intra-Latina/o variations in attitudes regarding gender.A small body of empirical studies has explored gender gaps in these attitudes among Latinas/os, with surprisingly equivocal results (Leaper and Valin 1996;Montoya 1996).Much of the scholarly attention in this area has centered on (a) indicators of acculturation such as generation, length of time in the United States, and language spoken at home and (b) structural factors such as education, income, and employment status.Briefly, the available evidence indicates that U.S. born Latinas/os or Latina/o immigrants who have spent significant time in the country tend to hold more egalitarian and less traditional gender ideologies than others (Parrado and Flippen 2005;Su et al. 2010).Indicators of acculturation such as speaking English (versus Spanish) at home have also been associated with more egalitarian gender ideologies (Kane 2000;Vasquez-Nuttall et al. 1987).However, some researchers have argued against the emphasis on acculturation since this focus caricatures and exaggerates the degree of male dominance in Latina/o culture, wherein micro-level gender relations are highly contested (e.g., (Baca Zinn 1980;Segura 1992)).In addition to acculturation, studies have shown that structural factors like education, employment, and economic resources are associated with greater gender egalitarianism among Latinas/os and other groups (Phinney and Flores 2002;Vasquez-Nuttall et al. 1987;Wilkie 1993).
Our study advances prior scholarship by drawing on critical insights from gender theory (e.g., (Cadge 2004;Connell 2009;Kimmel 2013;Lorber 2009;Ridgeway 2011;Sullins 2006)).Traditional gender ideologies involve different domains of belief that should be analyzed separately rather than combined together into a single index (Bartkowski and Shah 2014; see also (Bartkowski and Xu 2010)).It is particularly important to distinguish between three different ideological domains on which support for gender inequality often rests: (a) female domesticity, reflected in beliefs that women should focus principally or solely on 'private sphere' (household) pursuits while leaving the 'public sphere' (workplace and politics) to men; (b) gender essentialism, consisting of beliefs that women and men have very different 'natural' (even innate) capabilities such as women's putatively superior proclivity for emotional attachment and the nurturing of young children; and (c) patriarchy, composed of beliefs that men should serve as the authority figure, leader, or final decision-maker in key institutional settings, particularly in the home (i.e., husband 'headship').
Domain-specific analyses of gender ideology have received empirical substantiation, particularly where religion is concerned (Bartkowski and Shah 2014).For instance, theological conservatism and religious attendance predict gender essentialism but not beliefs related to separate spheres (Bartkowski and Xu 2010).Moreover, the social underpinnings of gender ideology sometimes work differently for religious men than for women (Bartkowski and Hempel 2009).Furthermore, qualitative research has revealed that ideological commitments to gender essentialism among religious adherents do not lead ineluctably to support for patriarchy or separate spheres (Bartkowski 2001;Mihelich and Storrs 2003).

Religion, Family, and Gender among Latinas/os
Apart from seeking to offer this theoretical advance, the current study also examines Latina/o gender ideologies with an eye toward religious influences.One notable omission from studies of gender ideology among Latinas/os has been the potential role of religion.Indeed, with few exceptions (Ellison et al. 2013;Hunt 1998Hunt , 2000Hunt , 2001)), investigators have given short shrift to the (a) interplay of traditional Catholicism and Latina/o culture in shaping family and gender norms among Latinas/os; (b) the relatively high levels of conventional religiosity among many Latina/o subgroups, and (c) the growth of Protestantism, especially conservative (i.e., fundamentalist, evangelical, and charismatic) forms, among Latinas/os in the United States and throughout Latin America.
Catholicism remains the dominant faith tradition among U.S. Latinas/os, with approximately 65 to 70 percent identifying as Catholic (Espinosa et al. 2005;Perl et al. 2006;Pew Research Center 2014), yet certainly not all who profess Catholicism attend mass regularly or partake in other church rites.Scholars argue that Catholicism has long been intertwined with Latina/o family and gender norms, reinforcing values such as familism and machismo (Diaz-Stevens 1994;Williams 1990).Indeed, critics have observed that the official Catholic Church embraces sharply gendered institutional roles, with priestly leadership roles reserved for males.There is, however, the empowering image of the Virgin of Guadalupe within popular Mexican-American Catholicism (Peña and Frehill 1998;Rodriguez 1994).Cultural images of the Virgin of Guadalupe, however, are diverse and complicated.She is a powerful symbol of Mexican identity.Yet, where gender is concerned, she has been associated with traditionalist ideals (motherhood, conventional feminine virtue) and with gender egalitarianism (feminism, social justice) (Campesino and Schwartz 2006).
It has been argued that 'the study of this religion is crucial for an understanding of all Hispanic peoples' (Espin 1994, p. 313).Such arguments raise the possibility that Latina/o Catholics may hold particularly traditional or patriarchal attitudes regarding gender roles compared with other Latinas/os.This body of research leads us to hypothesize the following: H1a: Catholics will be more inclined than conservative Protestant Latinas/os to stress (a) female domesticity, (b) gender essentialism, and (c) male headship within the family.
Despite the continued dominance of Catholicism within most sectors of the U.S. Latina/o population, evidence shows that, in recent years, the percentage of U.S. Latina/o Catholics has declined from 58 to 48 percent (Pew Research Center 2015) and that the percentage of Latina/o Protestants has markedly increased (e.g., (De la Torre 2008;Diaz-Stevens and Stevens-Arroyo 1998;Mulder et al. 2017)).Scholars estimate that roughly of 20 to 25 percent of U.S. Latinas/os now belong to Protestant churches, most of which are theologically conservative in orientation (Espinosa et al. 2005;Perl et al. 2006;Pew Forum on Religion & Public Life 2006).This shift to conservative Protestantism among Latinas/os may be partly explained by the renewalist movement, a fast-growing group of charismatics who hold beliefs in supernatural phenomena and prophetic revelations.This movement has had a much stronger impact on Latina/o versus non-Latina/o Christians (Heimlich 2007).Even so, conservative Protestantism seems to vary across Latinas/os of different countries of origin, with evidence suggesting that Puerto Ricans are more likely than Mexicans to identify with conservative Protestantism.However, across mainline and conservative Protestant churches altogether, Mexicans are the largest group, comprising over half of all Hispanic adherents (Pew Research Center 2014).
Previous studies indicate that Latinas/os belonging to conservative Protestant groups hold highly traditional views on a range of family-related values such as marriage, divorce, cohabitation, sex outside committed relationships (Ellison et al. 2013), abortion rights (Bartkowski et al. 2012), same-sex marriage (Ellison et al. 2011), and political ideology and partisanship (Kelly and Morgan 2008;Lee and Pachon 2007).Other research notes distinctions between converts and life-long adherents of conservative Protestantism, suggesting that those who convert are more liberal than life-long adherents, although this is true only for men.Among women, conversion is associated with increased traditionalist views (Hunt 2001).
Moreover, several studies have shown that, in the general U.S. population, members of conservative Protestant denominations tend to embrace more patriarchal gender attitudes than adherents of other faith traditions (e.g., (Hoffmann and Miller 1997;Peek et al. 1991)).This may partly explain why women from conservative Protestant denominations have lower levels of educational attainment, labor force participation, and earnings (Glass and Jacobs 2005;Lehrer 2009).It is reasonable to expect this similar support for patriarchal values among U.S. Latina/o conservative Protestants as well (Hunt 2001).This body of research suggests the following competing hypothesis, which runs directly counter to H1a: H1b: Conservative Protestants will be more inclined than Catholic Latinas/os to emphasize (a) female domesticity, (b) gender essentialism, and (c) male headship in the family.
Religious involvement or commitment, often measured in terms of attendance frequency, may also be linked to gender ideology (e.g., (Hertel and Hughes 1987)).Specifically, individuals who frequently attend religious services may be exposed to traditional conceptions of gender through sermons and other formal church pronouncements.These ideals may be reinforced via informal interactions with fellow church members in social settings sponsored by the religious community.Additionally, Sunday school classes, family ministries, and other educational contexts may convey patriarchal ideas that the roles of women are limited to homemaking and child-rearing, whereas the roles of men involve leadership positions both inside and outside of the home (Bartkowski 2001).Finally, religious groups may also convey patriarchal norms by valorizing gender traditionalism to laypersons within their communities in the assignment of domestic tasks such as cooking and child care to women while designating teaching and leadership positions to men.It is also possible that religious communities serve as 'reference groups,' whereby members conform to their norms by modeling desirable and acceptable lifestyles out of fear of informal sanctions (e.g., (Cochran et al. 2004)).
Frequency of religious attendance may also be a 'marker' of prior religious commitment.Viewed from that perspective, a link between attendance and gender traditionalism would reveal that more religiously committed persons hold more traditional gender ideologies, while less traditional persons might opt out of religion, perhaps due to its (real or perceived) social conservatism.Empirical studies have noted heightened levels of religious involvement among Latina/o Protestant converts (male and female) compared to their Catholic counterparts (Greeley 1994;Hunt 2001).Taken together, these strands of argument and evidence suggest the following study hypothesis: H2: Frequency of religious attendance will be associated with support for (a) female domesticity, (b) gender essentialism, and (c) male headship within the family.
In the general population, a belief in biblical literalism is linked with more traditional gender attitudes and practices (Ellison and Bartkowski 2002;Sherkat 2000).While a number of studies have demonstrated this pattern, the precise nature of this relationship is unclear.The Bible is a multi-vocal text that makes 'literal' reading difficult, if not impossible.Many researchers now recognize that 'literalism' actually refers to 'interpretive strategies' that are adopted by communities of scholars, theologians, pastors, and laypersons to denote readings and points of textual emphasis (Boone 1989;Malley 2004).For example, communities may emphasize readings involving sexuality, gender, family, etc. over other text that focuses on economic matters, assistance to the poor, and other topics (Sherkat and Ellison 1997).Individual laypersons often come to identify with 'literalist' pastors and congregations with whom they share general theological beliefs and social values.These laypersons may study supporting scriptural materials later, if at all.Thus, 'literalism' may reflect cultural boundaries, identities, and shared views on religious and social issues rather than reflecting an actual literal reading of the scriptural text itself (e.g., (Gallagher and Smith 1999;Sherkat and Ellison 1997)).
With respect to gender ideologies, there are scriptural readings that seem consistent with patriarchy and male headship such as those passages in the Bible instructing women to 'remain silent' in church and to 'submit' to their husbands 'in everything' (e.g., 1 Peter 2:11-12; Ephesians 5:22).Many denominations (primarily fundamentalist, evangelical, or charismatic Protestant) that endorse biblical literalism or inerrancy are the same ones that have maintained patriarchal organizational structures that deny women opportunities for formal leadership (Hoffmann and Bartkowski 2008).Researchers have identified strategies used by women to negotiate status in these groups (Bartkowski 2001;Denton 2004) in addition to empowerment resources like women's groups (e.g., (Brasher 1997;Griffith 2000)).Although these issues remain understudied among Latinas/os, one study links biblical literalism with support for male headship within the home (Ellison et al. 2013).This discussion suggests a final study hypothesis: H3: Biblical literalism will be associated with support for (a) female domesticity, (b) gender essentialism, and (c) male headship within the family.

Data and Measures
This study uses data from the National Survey of Religion and Family Life (NSRFL), a 2006 nationwide phone survey of approximately 2400 working-age adults (ages 18-59) in the contiguous United States.Latinas/os were oversampled (n = 801) and were interviewed by bilingual interviewers in Spanish or English.The NSRFL contains data on religious beliefs, affiliation, and practices, as well as attitudes regarding gender ideology and a number of other family-related issues.
Although the NSRFL response rates are low by traditional standards (36 percent, base sample, 41 percent Latina/o oversample), they are consistent with the response rates of other respected random digit dialing (RDD)-based surveys (e.g., (Council on Market and Opinion Research 2003;Edgell et al. 2006)).Growing evidence indicates that low response rates do not necessarily result in sampling bias (e.g., (Groves 2006)).In most cases, high response government surveys and low response RDD-based surveys demonstrate few substantial differences (Pew Research Center 2004).
From the 801 respondents in the Latina/o sample, 55 identified as mainline Protestants and subsequently excluded from our sample.Nine additional observations were excluded for item nonresponse to key gender ideology measures.Missing data for predictor variables were then handled using multiple imputation on Stata, yielding us an effective sample size of 737.We do not weight the data since we did not use the full Latina/o sample.

Dependent Variables
We used three dependent variables to tap attitudes towards different facets of traditional gender ideology.These variables were based on single items which gauged the extent to which respondents believe or agree with the following statements: 1.
'Preschool children are likely to suffer if their mother works;' 2.
'Women are better than men at taking care of young children;' 3.
'The husband should be the head of the family.'Consistent with the above theoretical arguments concerning the multidimensional nature of gender ideologies, these items measured three different sets of beliefs about gender: (a) female domesticity (such that mothers of young children should prioritize domestic pursuits); (b) gender essentialism (such that women are seen as 'naturally' more nurturing, in this case, more capable of caring for children than men), and (c) male headship (such that wives should defer to their husband's household authority).Agreement with each of these statements was scored on a four-point Likert scale, where 1 = strongly disagree; 2 = disagree; 3 = agree; and 4 = strongly agree.Higher scores for these variables indicated more traditional attitudes.Although we considered creating a multi-item index from the three gender ideology items, a low Cronbach's alpha score indicated insufficient correlations among items to justify this decision.Moreover, we have argued that current conceptualizations of gender emphasize the complex and multifaceted quality of gender ideologies.Consequently, there are sound theoretical reasons for examining a series of dependent variables to determine the social antecedents related to each of these ideological domains of gender (respective beliefs about female domesticity, gender essentialism, and male headship).For these reasons, each gender ideology item was analyzed separately using ordered logistic regression estimation techniques (Aiken and West 1991).

Independent Variables
Denomination was measured using dummy variables to identify Catholics, conservative Protestants (the vast majority of whom are fundamentalist, evangelical, or charismatic), and the religiously unaffiliated.Our comparison group consisted of Catholics because this is the affiliation of a significant majority of Latinas/os (Perl et al. 2006).The small numbers of respondents who did not answer this item or whose religious affiliation did not fit within the three broad categories were dropped from the analyses.An ordinal variable measured religious attendance using six categories, where 1 = never; 2 = a few times a year; 3 = once or twice a month; 4 = almost every week; 5 = once a week; and 6 = more than once a week.Our biblical literalism measure gauged the level of agreement with the statement: 'The Bible is the literal word of God and a true guide to faith and morality'.Response categories were scored on a four-point Likert scale, where 1 = strongly disagree; 2 = somewhat disagree; 3 = somewhat agree; and 4 = strongly agree.Items were reverse-coded so that higher scores reflected stronger endorsements.

Control Variables
We included additional variables to adjust for factors that may confound the associations between religion and gender ideology.Three dummies were used to explore the differences in respondent culture.The first measured interview language, used as a proxy for acculturation, where 1 = Spanish interview and 0 = English interview.The second variable measured citizenship status, where 1 = non-citizen and a 0 = U.S. citizen.The third variable measured national origin, where 1 = Mexican American and 0 = 'other' Latina/o (e.g., (Puerto Rican, Cuban American, Dominican, etc.)).

Analytic Strategy
After presenting unweighted sample characteristics, we display four ordinal logistic or stereotype regression models estimating the net effects of religious variables and covariates on three domains of gender ideology.Model 1 included religious affiliation and controls.Model 2 substituted religious attendance for affiliation.Model 3 replaced attendance with biblical literalism, and in the full model, Model 4, all religious variables were considered simultaneously, along with the controls.This approach enabled us to see the net effects of each dimension of religious involvement on the respective domains of gender ideology (female domesticity, gender essentialism, and male headship).When the proportional odds assumption was violated for the ordinal logistic regression models, stereotype regression was used, which is free of this assumption.

Descriptive Statistics
Table 1 displays unweighted sample characteristics.Approximately 7% of respondents were conservative Protestants, 79% were Catholic, and the remaining 14% were religiously unaffiliated.The average Latina/o respondent in the NSRFL reported attending religious services roughly once or twice a month and indicated relatively strong support for biblical literalism.Overall, the respondents also expressed tepid agreement with the items tapping traditional gender ideology.In terms of demographic characteristics, the average Latina/o respondent was 37 years old, with some college education and a household income of $31,000.Over half (52%) were employed.Roughly 64% were Mexican American, while 56% were non-citizens and 51% interviewed in English.Two thirds were women, while 56% were married, with an average of 2.1 children in the home.

Estimated Net Effects of Religious Variables
Table 2 displays the results from four ordered logistic regression models that estimated the net effects of religious variables and controls on attitudes concerning working mothers with young children (support for female domesticity).Model 1 indicated no significant differences between conservative Protestants and Catholics (OR = 1.113; not significant) when it comes to the belief that 'Preschool children are likely to suffer if their mother works'.This finding offers no support for H1a or H1b.Turning now to Model 2, religious attendance was associated with support for female domesticity (OR = 1.134; p < 0.01), which is consistent with H2.According to Model 3, biblical literalism was not associated with concern for the children of working mothers (OR = 1.157; not significant), offering no support for H3.When all religion variables were entered simultaneously in Model 4, only religious attendance remained significantly associated with this facet of gender ideology (OR = 1.112; p < 0.05).Next, we turn to a similar sequence of models in Table 3, which examined support for the view that women are better suited for child care than men (gender essentialism).Here we found no meaningful variations in support for this view by denominational affiliation, religious attendance, or biblical literalism in any model.These patterns yielded no support for H1a, H1b, H2, or H3, leading us to reject all hypotheses with respect to this dependent variable.Finally, Table 4 displays stereotype regression models estimating the net effects of religious variables and covariates on support for male headship (patriarchy) within the home (i.e., the belief that the husband should be the head of the family).The results presented here diverge markedly from the modest findings reported above.Consistent with H1b, Model 1 revealed that the odds to strongly disagree, disagree, or agree with this view are approximately 74%, 70%, and 60% lower [OR (SD vs. SA) = e (−1×1.358)= 0.26, OR (D vs. SA) = e (−0.883×1.358)= 0.30, and OR (A vs. SA) = e (−0.679×1.358)= 0.40; p < 0.001], respectively, for conservative Protestants than for Catholics (relative to strongly agreeing).As anticipated by H2, Model 2 revealed a positive association between the frequency of religious attendance and support for male headship.That is, for each unit increase in religious attendance, the odds to strongly disagree, disagree, or agree with male headship decrease compared with strongly agreeing [OR (SD vs. SA) = e (−1×0.268)= 0.76, OR (D vs. SA) = e (−0.795×0.268)= 0.81, and OR (A vs. SA) = e (−0.636×0.268)= 0.84; p < 0.001].In Model 3, agreement with biblical literalism was linked with an approval of male headship [OR (SD vs. SA) = e (−1×0.700)= 0.50, OR (D vs. SA) = e (−0.807×0.700)= 0.57, OR (A vs. SA) = e (−0.594×0.700)= 0.66; p < 0.001], a pattern which supported H3.When all the religion variables were included in Model 4, conservative Protestants remained more inclined to endorse male headship than Catholics [OR (SD vs. SA) = e (−1×1.149)= 0.32, OR (D vs. SA) = e (−0.840×1.149)= 0.38, OR (A vs. SA) = e (−0.611×1.149)= 0.50; p < 0.001].The findings regarding religious attendance [OR (SD vs. SA) = e (−1×0.189)= 0.83, OR (D vs. SA) = e (−0.840×0.189)= 0.85, OR (A vs. SA) = e (−0.611×0.189)= 0.89; p < 0.01] and biblical literalism [OR (SD vs. SA) = e (−1×0.613)= 0.54, OR (D vs. SA) = e (−0.840×0.613)= 0.60, OR (A vs. SA) = e (−0.611×0.613)= 0.69; p < 0.001] remained statistically significant in the full regression model.

Estimated Net Effects of Covariates
Although not the primary focus of our study, the estimated net effects of several covariates warrant brief discussion.First, consistent with previous theory and research (Davis and Greenstein 2009;Kane 2000), education was consistently associated with more progressive gender ideologies.By contrast, household income, another SES indicator, had no bearing on any facets of gender ideology considered here.Second, women expressed more progressive views when compared to men concerning gender essentialism and male headship, with the exception of female domesticity.Third, respondents who interviewed in Spanish (versus in English) strongly endorsed more traditional gender ideologies, a finding consistent with literature linking cultural assimilation with gender traditionalism.Overall, this pattern is broadly consistent with previous studies highlighting the role of factors such as SES, generation, and acculturation with Latina/o gender and family patterns (e.g., (Raley et al. 2004;Su et al. 2010)).Few other variables emerged as significant predictors of gender ideology among this sample of Latinas/os.

Ancillary Analyses
In addition to the main analyses reported above, we examined three possible sets of contingent effects by adding zero-centered multiplicative interaction terms to the full models presented in Tables 2-4.First, we considered whether the associations between religious factors and gender ideology differed for men and women.In addition, given that the language of the interview (Spanish versus English) was used as a very rough indicator for the degree of acculturation, we investigated whether the role of religion was stronger among those who completed the interview in English as opposed to those who interviewed in Spanish.Finally, we explored whether the estimated net effects of religion varied with the respondents' education level.There was no evidence of any significant interactive associations.To the contrary, the main patterns of interest appeared quite robust across these key Latina/o population subgroups.

Discussion
Although prior research has examined religious variations in gender ideology among Americans in general, few studies have addressed this topic among U.S. Latinas/os.This oversight is surprising in light of (a) the extensive evidence of religious variations in gender ideology within the overall U.S. population; (b) the rapid growth of the U.S. Latina/o population; (c) the increasing religious diversification among Latinas/os, fueled by the expansion of (primarily conservative) Protestantism; and (d) the longstanding interest in Latina/o cultural values regarding gender and family, most notably, familism and machismo.Analyzing data from a sample of working-age Latinas/os in the United States, our study has explored the associations between multiple dimensions of religious involvement (denominational affiliation, attendance, and biblical literalism) and three distinct facets of traditionalist gender ideology: concerns about possible consequences of women's labor force participation (female domesticity), beliefs that women are better suited for child care (gender essentialism), and beliefs about male headship in the household (patriarchy).Current developments in gender theory and supporting empirical scholarship warrant this approach.Several findings merit discussion.
One overall pattern is clear; religious factors loomed large for the most symbolic gender ideology variable, the belief in male headship (patriarchy).Consistent with one key hypothesis, Latina/o conservative Protestants, who tend to be primarily fundamentalist, evangelical, or charismatic were much more supportive of patriarchy than were other Latinas/os.Religious attendance and level of agreement with biblical literalism were also strongly predictive of support for the male headship ideal.Regarding female domesticity, religious attendance was the only variable that strongly predicted concern about the harmful effects of women's employment on children.Finally, regarding women's putative penchant for child-rearing (gender essentialism), religion had little bearing on this domain-specific gender ideology item.Prior research conducted with white conservative Protestants has revealed a pattern similar to that found here.Attitudinal support for male headship of the family is a method that conservative Protestants employ for symbolically distinguishing themselves from the cultural mainstream.However, because attitudinal support for husband headship is largely symbolic, it quickly gives way to (more) egalitarian convictions and practices related to other forms of gender (Denton 2004;Gallagher and Smith 1999).For this reason, we see few significant effects with respect to domesticity and child-rearing.Assimilation (indicated by interview language) was the strongest predictor of (a) concern about the implications of women's labor force activity on the well-being of children and (b) the essentialist belief that women are better suited for child care than men.
Moreover, Latina/o Catholics exhibited little enthusiasm for gender traditionalism.There were no significant differences between Catholics and religiously unaffiliated Latinas/os on any of the gender ideology items considered here.This finding contradicts longstanding images of Latina/o Catholicism as a bastion of cultural conservatism and a carrier of traditional Latina/o values, presumably including familism and machismo.To the extent that male headship within the home is consistent with the ethos of machismo, it is actually conservative Protestantism that appears to be the carrier of this traditional value orientation today.Moreover, evidence of Protestant distinctiveness persisted even with statistical controls for biblical literalism.Future research should investigate the underlying reasons for conservative Protestant subcultural distinctiveness with respect to male headship.Is this conviction the product of specific congregational factors such as sermon content, recommended gender practices, religious education and pastoral materials, religious media, or other within-group influences?What is the possible role of selectivity, (i.e., what kinds of persons choose to remain or become conservative Protestant and what is the place of gender and family factors in their decisions)?Interestingly, in contrast to findings from U.S. general population samples, religious attendance, treated here as an indicator of religious involvement or commitment, was unrelated to gender ideology in virtually all models.Thus, although much has been said about a broad religious-secular divide in the general public opinion on a range of issues, religious beliefs and subcultural differences (as opposed to the degree of religious involvement or commitment) were more closely linked with gender ideology among Latinas/os.These findings lend further credence to warn against gender ideology indexes that combine measures gauging distinct facets of gender inequality (Bartkowski and Xu 2010).Beliefs about female domesticity or separate spheres, gender essentialism, and patriarchy have been shown to have different social and even religious correlates and antecedents.
Two additional points are especially germane to future research on the role of religion in gender and family life.First, the estimated net effects of conservative Protestant affiliation and biblical literalism on support for male headship within the home (patriarchy) were largely independent of one another.This finding demonstrates that affiliation and belief are not interchangeable indicators of Protestant conservatism.To the contrary, previous work has revealed that significant numbers of Latina/o Catholics expressed agreement with literalism in the NSRFL (Ellison et al. 2013), thereby raising the possibility that this item partly taps the strength of religious conviction as well as conservative theological particularism within this population.We note that the wording of the biblical literalism item in the NSRFL differs slightly from the wording of similar items in other major surveys such as the National Opinion Research Center (NORC) General Social Survey.Additional studies, perhaps using qualitative interviews (e.g., (Peña and Frehill 1998)), should clarify the way(s) in which Latina/o respondents interpret and answer this and other survey items on religion.Such research, in turn, will help determine the impact, if any, of question wording on the pattern of results reported here.More generally, the finding that biblical literalism is not a consistent marker of gender traditionalism among Latinas/os, in contrast to the general population, adds to a small but growing body of evidence that 'literalism' and 'inerrancy' (and other facets of religiosity) play different roles across racially, ethnically, and culturally distinctive groups (McDaniel and Ellison 2008;Read and Eagle 2011).
Second, our finding regarding conservative Protestantism and advocacy for male headship adds to a growing body of literature on Latina/o Protestant distinctiveness.Previous studies have reported that, on average, conservative Protestants (a) attend services and pray more often than other Latinas/os (e.g., (Diaz-Stevens and Stevens-Arroyo 1998)) and (b) hold more conservative attitudes on abortion (Bartkowski et al. 2012), same-sex marriage (Ellison et al. 2011), marriage, divorce, sexuality (Ellison et al. 2013), political ideology (Kelly and Morgan 2008), and partisanship (Lee and Pachon 2007).Thus, our study indicates yet another critical dimension through which conservative Protestant Latinas/os distinguish themselves from their Catholic counterparts, namely, attitudinal support for male headship.At the same time, we are not suggesting that the actual practice of patriarchy in conservative Protestant Latina/o families follows these attitudinal differences in a lockstep fashion.As noted, white conservative Protestants often symbolically embrace the language of headship while practicing relatively egalitarian decision-making (Denton 2004;Gallagher and Smith 1999).More research is clearly needed to determine if this paradox holds among Latina/o conservative Protestants.
This study was characterized by several limitations.First, as is often the case, our data were cross-sectional, thereby precluding any conclusive claims about the causal effects of religion on gender ideology.Instead, we have established patterns and associations that could be profitably investigated using longitudinal data in the future.Second, although our sample size was adequate for the purpose of this study, it would be desirable to have a larger sample that would permit more fine-grained analyses of variations by national-origin groups and other potential sources of diversity within the Latina/o population.The NSRFL is also a sample of only the working-age (18-59 year-old) Latina/o population, and it would be useful to explore the associations between religious factors and gender ideology across the full age spectrum as well.Third, as noted earlier, the response rate of the survey was rather low, although it compares favorably with the response rates of many other large-scale telephone surveys conducted in recent years, particularly those involving Latinas/os.To be sure, a growing literature has demonstrated that low response rates need not imply biased results (Groves 2006;Massey and Tourangeau 2013;Peytchev 2013).Nevertheless, it would be ideal to have a higher response rate in the future.These limitations notwithstanding, this research has cast fresh light on an important and neglected topic.

Conclusions
This investigation aimed to fill a noteworthy gap in the research on the religious sources of gender inequality among racial-ethnic minorities.We focused on the relationship that denominational affiliation (generally Catholic versus conservative Protestant), worship service attendance, and biblical literalism have with gender ideology.The last of these religious variables was included because of the substantial inroads that conservative Protestantism has made within the American Latina/o population.Our findings have revealed that the associations between religious factors and gender ideology among Latinas/os are complex and nuanced, perhaps even more so than in the overall population.Drawing insights from gender theory, we focused on three important facets of gender ideology: female domesticity, gender essentialism, and patriarchy.It is the last of these, namely, patriarchy that is most robustly linked with religious factors.This finding suggests that, while the practice of machismo may be tempered by conversion to conservative Protestantism, the symbolic importance of men's leadership in the home is amplified within this particular faith tradition.
Further research is needed to explore religious differences, if any, in gender practices within the home (e.g., the division of household labor, child care and child-rearing, household decision-making) and beyond.Additional quantitative scholarship could explore the magnitude of any distinctions in the practice of gender within Latina/o Catholic households versus Latina/o conservative Protestant families, particularly given prior survey research that demonstrates how symbolic leadership does not always support distinctive gender behaviors (Denton 2004).At the same time, qualitative research could examine if gender negotiations are different in such families than in white conservative Protestant homes (Bartkowski 2001).Such work would complement the present study and enrich our understanding of the relationships between religion, gender ideologies, and gender practices among Latinas/os in the contemporary United States.

Table 1 .
Sample Characteristics after Multiple Imputation.

Table 2 .
Odds Coefficients of Ordered Logit Models Predicting Female Domesticity: 'Preschool Children Are Likely to Suffer If Their Mother Works'.

Table 3 .
Odds Coefficients of Ordered Logit Models Predicting Gender Essentialism: 'Women Are Better Than Men at Taking Care of Young Children'.

Table 4 .
Logit Coefficients of Stereotype Regression Models Predicting Male Headship: 'The Husband Should Be the Head of the Family'.