Traditional feminine gender roles, alcohol use, and protective behavioral strategies among Latina college students

Abstract

Objective:

Studies conducted with Latina/o populations suggest there are stringent gendered norms surrounding drinking behavior, but research examining alcohol use among Latina college students in the context of marianismo, a traditional Latina feminine gender role, is scarce.

Participants and Methods:

405 Latina undergraduates completed a web-based study of sociocultural factors and alcohol use. Path analyses assessed the relation between marianismo, alcohol use, and alcohol-related protective behavioral strategies (PBS).

Results:

A dimension of marianismo characterized by purity was related to (a) reduced alcohol use in general, and (b) PBS designed to modify the manner of drinking among drinkers. In turn, this dimension of PBS related to less frequently experiencing drunkenness.

Conclusions:

These findings support the notion that enacting some dimensions marianismo among Latina college students may limit general alcohol use (directly) and high-risk drinking (indirectly through use of PBS), while enacting other dimensions may place Latinas at risk of alcohol misuse.

Introduction

National reports indicate that alcohol use is prevalent among college women.¹ Research shows that women will attain higher blood alcohol concentration levels and will become intoxicated faster than men during the same drinking period.² Also, sexual victimization and other negative sexual consequences have been linked to alcohol use among college women.³

Past U.S. trends suggest men consume more alcohol than women, although the gender gap is narrowing.⁴ Gendered societal expectations for drinking may influence this gender gap. In general, women perceive that society discourages alcohol use for women more than men.⁵ Some suggest the societal prescriptions for drinking behavior may be more stringent in some cultures than others. For instance, scholars suggest that norms discouraging health-risk behaviors among women (while encouraging the same behavior among men) are particularly prevalent among many Latina/o cultures.⁶ Indeed, studies examining alcohol and other forms of substance use in relation to traditional Latina/o gender roles indicate conformity to certain Latino gender roles (i.e., traditional machismo) is a risk factor for hazardous alcohol use (e.g., binge drinking among men),⁷ while other aspects of Latina gender roles (i.e., marianismo) is protective against general substance use.⁸

An individual’s engagement in alcohol use is influenced by norms transmitted through a person’s immediate and distal environment.⁹ Accordingly, the salience of alcohol-related norms may differ according to context.⁹ For many Latinas, attending college may provide a novel opportunity to engage with newly intersecting norms and identities. That is, some internalized traditional norms may be challenged by norms encountered in the college environment. Thus, if Latina/o heritage culture discourages drinking among women, the internalization of these norms may limit drinking behavior (as noted by Lee et al¹⁰). However, given the prevalence of alcohol use among college women,¹ a Latina student may conform to the drinking norms of the college environment. In that case, it is still possible that the heritage cultural norms, such as some aspects of marianismo, can limit heavy drinking. Therefore, it is important to go beyond examining the link between heritage cultural norms and general alcohol use, and to also explore how these norms may be enacted should alcohol be consumed.

Marianismo

Within a social constructionist framework, scholars argue that cultural norms provide a rubric for what it means to be ‘men’ and ‘women’. Rather than existing somewhere within the individual, gender is instead an agentic enactment in response to perceived societal expectations and stereotypes.¹¹ Thus, if society prescribes different behaviors for men versus women, engagement in these behaviors can be conceived of as ‘performing gender’ (as noted by Wilkinson et al¹²). When it comes to alcohol use, researchers speculate that, in general, Latina/o men – but not women – are encouraged to engage in this behavior¹³.

The label commonly used to describe the Latina feminine gender role is marianismo. Historically, researchers have described marianismo as rooted in conceptualizations of the idyllic Virgin Mary.¹⁴ Contemporary research demonstrates that marianismo is a complex system of Latina gender role prescriptions that are infused with Latina/o cultural values (see Castillo et al¹⁵. Under the guidelines of marianismo, among other things, a Latina is expected to be chaste, pure, deferent, and serve as the spiritual backbone for her family. Using a sample of Latina college students, Castillo and colleagues¹⁵ created the Marianismo Beliefs Scale (MBS). The authors found marianismo was best characterized by five dimensions. Two of these suggest that a Latina should be a unifying form of strength for her family in general (i.e., ‘Family Pillar’) and a caretaker for their spirituality (i.e., ‘Spiritual Pillar’). Two others suggest that Latinas should be subordinate, particularly to men (i.e., ‘Subordinate to Others’), and exercise self-repression (i.e., ‘Silencing Self to maintain harmony’). The fifth indicates that Latinas should also behave in a manner that would be considered pure (i.e., ‘Virtuous and Chaste’).

Marianismo and alcohol use

The link between marianismo and drinking behavior is not well understood, due in part to limited research on this topic. A recent literature review highlights the mixed nature of the associations between marianismo-like constructs and alcohol use.²¹ For example, researchers found that traditional gender role (TGR) beliefs that emphasized feminine subservience and purity was inversely related to alcohol use for Latina adolescents.¹⁶ Another study of Latina adolescents indicated a negative association between a feminine identity characterized by affective femininity (e.g., nurturance, empathy) and drinking frequency. Supporting the notion that elements of marianismo may be the salient gender role in this circumstance, the same study found that affective femininity was particularly protective against high risk drinking for Latinas who were not as oriented to mainstream U.S. culture.¹⁷ More recently, a study⁷ examined marianismo beliefs and alcohol use among Latina college students and collapsed the five dimensions of the MBS¹⁵ into two, similar to a previous study examining the marianismo construct in relation to academic outcomes.⁴⁸ The first dimension, positive marianismo, was comprised of dimensions characterized by virtue, familial strength, and spiritual strength. The second dimension, negative marianismo, was comprised of dimensions characterized by self-repression and subordination. The authors found no main effects between the marianismo dimension and either general alcohol use or binge drinking, however.

The reasons for the null main effects from Perrotte et al⁷ are not entirely clear, but it is possible that only some dimensions of marianismo are related to alcohol use, while others are either unrelated or indirectly related through other variables. That is, collapsing across dimensions of marianismo may mask important nuances of each dimension that may differentially relate to alcohol use. For example, several aspects of marianismo may be linked to distress.^18,19 Two dimensions correspond with Latinas being responsible for the happiness and spiritual growth of the family (i.e., Family Pillar and Spiritual Pillar). Two other dimensions correspond with repressing the outward expression of feelings (i.e., Silencing Self) and behaving in ways that are consistent with subordination (i.e., Subordinate to Others). It is possible that Latinas enacting some of these aspects of marianismo may consequently experience distress, which drinking alcohol may help ease. Conversely, the notion that a Latina should be ‘pure’ may more directly relate to decreased alcohol use to the extent that drinking is inconsistent with the values of purity. Indeed, research suggests a dimension of marianismo encompassing traits such as virtue and purity (i.e., ‘Virtuous and Chaste,’¹⁵) may be inversely related to substance use among Latina adolescents.⁸ The contrasting findings across studies^7,8 present evidence that the various aspects of marianismo may be distinctly related to drinking behavior, and speaks to the need for more research exploring the nature of the relation between all dimensions of marianismo and alcohol use. Hence, the current study examined the unique relations between each of the five dimensions of marianismo and alcohol-related outcomes.

While findings from the literature support the notion that some aspects of TGR for Latinas could be inversely related to alcohol use, it is still not clear how and why this may be. It is also important to consider that college-aged individuals are exploring their identity,²⁰ which may coincide with using substances such as alcohol. Thus, it is conceivable that Latina students who choose to drink but remain committed to some of the tenets of marianismo may rely on protective behavioral strategies (described in the next section) to decrease the likelihood that they will experience negative consequences associated with heavy drinking. In doing so, a Latina may still choose to enact marianismo while navigating the college environment.

Protective Behavioral Strategies (PBS)

Defined as cognitive-behavioral strategies one can use before, during, and after a drinking episode to limit alcohol consumption and related consequences, PBS have been associated with reduced drinking and alcohol problems among college students.²² The Protective Behavioral Strategies Scale PBSS,²³ specifies three types of strategies: those that limit serious alcohol-related harms (e.g., “Use a designated driver”), those that help a person stop or limit the amount of alcohol they consume (e.g., “Determine not to exceed a set number of drinks”), and strategies that help one modify how they drink alcohol so as to limit consumption and associated problems (e.g., “Avoid drinking games”). Strategies to help one modify the way they drink appear to be most effective in reducing alcohol use and negative consequences.^24,25

Research indicates college women tend to use more PBS than college men (e.g., Pearson²²), and evidence suggests PBS is more protective for women than men,²⁶ though more research is needed. Although some research indicates race/ethnicity are associated with PBS use and the extent to which PBS are protective of alcohol use and negative consequences,^27,28 PBS and alcohol use research that focuses specifically on Latina college students has not been conducted.

In addition to being examined in association with various drinking behaviors and alcohol-related problems, PBS have been examined as mechanisms linking various risk factors and alcohol outcomes. While PBS have not been examined as a mediator of marianismo and alcohol outcomes, previous research indicates women may engage in other safety-related activities when drinking in order to reduce the likelihood they are viewed negatively due to evident intoxication.²⁹ Given this along with evidence that PBS may link more distal predictors to alcohol outcomes, we hypothesize that PBS may account for a portion of the relation between marianismo and drunkenness among Latina college women.

Current Study

The current study examined the links between marianismo and general alcohol use, and links between marianismo, PBS, and drunkenness among Latina college women (Figure 1). Theory and research suggest engaging in alcohol use would be counter to the enactment of some dimensions of marianismo, but perhaps not others; however, research in this area is limited. Thus, our hypotheses are rooted in conceptualizing health-risk behavior (and the mitigation of such behavior) as a performance of aspects of TGR rooted in concepts of marianismo. First, we expect the dimension of marianismo related to purity will be inversely related to both general alcohol use (i.e., frequency and quantity) and high-risk alcohol use (i.e., drunkenness among drinkers) as it is the marianismo dimension that is conceptually most directly related to drinking behavior. As mentioned previously, the potential link between other dimensions of marianismo and alcohol related behaviors is not as evident, therefore the respective portion of the analysis is more exploratory. Since Latina college women may be negotiating between traditional values and those of the mainstream college environment, we posit that drinking-related aspects of marianismo may be enacted even among drinkers. Specifically, we expect that marianismo will be positively related to PBS that inhibit excessive alcohol intake (hypothesis 2), which in turn will be related to reduced likelihood of experiencing drunkenness (hypothesis 3).

Figure 1.

Conceptual model of the link between marianismo, protective behavioral strategies, and alcohol use.

Methods

Participants and procedure

These data come from an IRB approved study at a southwest university examining alcohol use among 584 Latina/o undergraduate men and women (69% female). Students were recruited during the fall of 2016 through Introduction to Psychology classes to participate in a 30-minute web-based survey in fulfillment of course requirements. For this study, participants identified as female and Hispanic or Latina, were between 18–25, and unmarried. Four hundred and five Latina students met these criteria (mean age = 18.65, SD = .10; 65.9% Mexican or Mexican American).

Measures

Marianismo

The Marianismo Beliefs Scale (MBS¹⁵) measured 5 dimensions of marianismo on a 7-point scale (1 = strongly disagree). Virtuous and Chaste contained 5 items (e.g., “A Latina should be pure,” and “A Latina should remain a virgin until marriage”), Family Pillar contained 5 items (e.g., “A Latina must keep the family unified,” and “A Latina must be a source of strength for her family”), Spiritual Pillar dimension contained 3 items (e.g., “A Latina must be the spiritual leader of the family,” and “A Latina is responsible for the spiritual growth of the family”), Subordinate to Others contained 5 items (e.g., “A Latina should avoid saying no to people,” and “A Latina should do anything a male in the family asks her to do”), and Silencing Self contained 6 items (e.g., “A Latina should feel guilty about telling people what she needs,” and “A Latina should be forgiving in all aspects”). Averages were calculated within each subscale; higher scores indicated higher endorsement of the respective dimension. The MBS was developed using a heterogeneous sample of Latina undergraduate students and demonstrated strong internal consistency within dimensions during development (α across dimensions ranging from .77 - .85). ¹⁵ Likewise, internal consistency of the marianismo dimensions for the current study was supported (Virtuous and Chaste α = .78, Family Pillar α = .82, Spiritual Pillar α = .85, Subordinate to Others α = .72, Silencing Self α = .70).

Protective Behavioral Strategies (PBS)

The Protective Behavioral Strategies Scale²³ measured PBS. All items were rated on a 6-point Likert scale (1 = Never, 6 = Always). Two subscales were used for this study. The first, ‘Stopping/Limiting Drinking’, has five items (e.g., “drink water while drinking alcohol”). The second, ‘Manner of Drinking’, asks participants about their engagement in high-risk drinking behaviors (e.g., “drink shots of liquor”) and regulation/avoidance of high-risk behaviors (e.g., “drink slowly, rather than gulp or chug”) along 7 items. Previous studies with college samples indicate both Stopping/Limiting Drinking and Manner of Drinking are internally consistent (α = .82 and .74, respectively).²³ The present study also found these scales to have good internal consistency (Stopping/Limiting Drinking α = .86; Manner of drinking α = .74).

General alcohol use

General alcohol use was measured using a modified Daily Drinking Questionnaire, Revised (DDQ-R³²), which has been deemed a reliable and valuable assessment tool for drinking behavior among college students. Participants reported the frequency of drinking on each day of the week for the past 30 days. These scores were summed to yield a total estimate of alcohol frequency for the past month. Participants additionally estimated the average quantity of alcoholic drinks they consumed on each day of the week in the past month. These scores were multiplied by the reported daily frequencies to yield a total quantity of estimated typical alcohol consumption the month. All items assessing quantity were presented with a chart depicting standardized drink portions.³³

Drunkenness

Modeled after previous research,³⁴ participants reported how often during the past 30 days they drank to the point of feeling “drunk (i.e., more than just a little high, lightheaded, or ‘buzzed’)” using whole numbers. Given the preponderance of zeros for the present analysis, responses were dichotomized to be interpreted as never experiencing drunkenness (0) or experiencing at least one episode of drunkenness (1) during the past 30 days.

Covariates

Time since beginning college was a covariate in all models pertaining to alcohol use because research indicates that alcohol use tends to increase over the course of college.³⁰ Scores for time since beginning college ranged from 1 (less than 6 months) to 7 (5 or more years). We also included relationship status as a covariate in the analysis of PBS given the variable’s relation to health regulatory behaviors.³¹ Participants reported being ‘single with no attachments’ (1), ‘casually dating one or more people’ (2), or ‘exclusively dating a single person’ (3).

Data analytic approach

Data were analyzed using Mplus version 8. During preliminary analyses we examined normality, correlations, and descriptive statistics for each variable. Inattentive responses were identified using established techniques (i.e., asking participants if they were attentive during the study and if their responses should be used for data collection)³⁵ and were omitted during primary analyses. All analyses employed maximum likelihood estimation.

The first model tested the relation between marianismo and general alcohol use among the entire sample. As is customary with substance use data (see Atkins et al³⁶), we fit a zero-inflated model in which each general alcohol use measure was regressed onto each dimension of marianismo. The zero-inflated analysis allowed us to assess how marianismo related to membership in two separate classes of zeros (i.e., structural zeros, or, never drinkers, and those who may drink but did not during the 30-day self-report window in the present study) using a logistic regression equation. The same model also assessed how dimensions of marianismo related to the count distribution of each general alcohol use measure.³⁶

A path analysis using only participants who reported past 30-day drinking assessed the relation between marianismo, PBS, and drunkenness. In this model, Stopping/Limiting Drinking and Manner of Drinking were each regressed onto each dimension of marianismo. Also, drunkenness was regressed onto both PBS subscales and onto each marianismo dimension. Since the drunkenness variable was dichotomous and the properties of a*b cannot be generalized to endogenous variables without a continuous distribution, indirect effects were tested using a counterfactual framework, which assesses the total effect, total natural indirect effect, and pure natural direct effect.³⁷ However, indirect effect models in Mplus using this framework require that the model cannot include multiple mediating variables. Thus, models assessing the total natural indirect effect of each of the two PBS subscales on the relation between marianismo and drunkenness were analyzed separately.

Results

Preliminary and descriptive analyses

Descriptive statistics suggested the dimensions of marianismo and PBS were normally distributed. See Table 1 for sample means, reliabilities, and correlations between marianismo and PBS. Mean comparisons on marianismo subscales across drinkers and non-drinkers indicated the means were not significantly different, thus the means for drinkers are displayed for a more parsimonious presentation. The marianismo dimensions were highly correlated with each other. Data showed a correlation between Virtuous and Chaste and Manner of Drinking, but no other preliminary relations emerged between marianismo dimensions and PBS.

Table 1.

Descriptive statistics and zero-order correlations among independent variables

		Mean(SD) Drinkers	1	2	3	4	5	6	7	8	9
1	TSBC	2.03(1.51)	-
2	Relationship Status	2.13(.92)	−.05	-
3	Virtuous and Chaste	4.29(1.32)	.05	.06	.78
4	Family Pillar	5.28(1.18)	−.01	.06	.57***	.82
5	Spiritual Pillar	3.57(1.57)	.06	.11†	.66***	.59***	.85
6	Subordinate to Others	2.30(1.13)	.08	.14*	.51***	.25***	.49***	.72
7	Silencing Self	2.42(.97)	−.01	.16*	.51***	.25***	.47***	.65***	.70
8	Stopping/Limiting Drinking	3.84(1.21)	.02	.03	.10	.00	.02	.14*	.07	.86
9	Manner of Drinking	3.57(1.10)	−.00	.19**	.16*	.02	.10	.13†	.10	.60***	.74

Note. Data include only past 30-day drinkers (n=222). Reliabilities in bold on diagonal. TSBC = time since beginning college.

^***p<.001,

^**p< .01,

^*p<.05,

^†p<.10.

Most participants reported attending college a year or less (72%) and a GPA of 2.6 or higher (69.3%). Most reported being in an exclusive relationship (50.1%) or single with no attachments (40.7%). Also, 222 (55%) reported engaging in any episodes of drinking during the past month. Among those who reported drinking in the past month, 91 (41%) reported never experiencing drunkenness, 48 (21.6%) reported one occasion on which they experienced drunkenness, 35 (15.8%) reported two occasions, 16 (7.2%) reported three occasions, 13 (5.6%) reported four occasions, and 18 (8.1%) reported five or more occasions of drunkenness. Regarding PBS, very few current drinkers reported never using Stopping/Limiting Drinking (n = 2) or Manner of Drinking (n = 5) strategies.

Primary analyses

After accounting for inattentive responses (n = 49) and missing data, the analytic sample was 337 drinkers and non-drinkers (first model) and 212 drinkers (second model and subsequent test of indirect effects).

Marianismo and general alcohol use (Table 2)

Table 2.

The Relation Between Marianismo and General Alcohol Use

	Frequency
	Zero-inflation				Count
Variable	B	SE	LLCI	ULCI	B	SE	LLCI	ULCI
Virtuous and Chaste	.07	.13	−.19	0.33	−.01	.08	−.16	.15
Family Pillar	−.04	.14	−.31	.22	−.01	.07	−.14	.13
Spiritual Pillar	−.10	.11	−.32	.13	.01	.06	−.10	.12
Subordinate to Others	−.30†	.16	−.61	.00	.02	.07	−.11	.16
Silencing Self	.33†	.19	−.05	.70	−.08	.09	−.25	.10
	Quantity
	Zero-inflation				Count
Variable	B	SE	LLCI	ULCI	B	SE	LLCI	ULCI
Virtuous and Chaste	.05	.13	−.20	.31	−.23*	.10	−.42	−.03
Family Pillar	−.06	.13	−.33	.20	−.02	.10	−.21	.17
Spiritual Pillar	−.11	.11	−.33	.11	.06	.08	−.10	.21
Subordinate to Others	−.27†	.15	−.57	.03	.03	.09	−.15	.20
Silencing Self	.39*	.19	.02	.76	.03	.12	−.21	.26

Note. Loglikelihood: −1189.95, AIC: 2439.91, BIC: 2554.51; LLCI = 95% lower limit confidence interval. ULCI = 95% upper limit confidence interval.

^***p< .001.

^**p< .01,

^*p< .05,

^†p< .10.

Both Subordinate to Others (OR = .74) and Silencing Self (OR = 1.39) were marginally related to membership in the structural zeros (never drinkers) of drinking frequency. None of the marianismo dimensions were related to the count portion of drinking frequency, however. Similarly, when assessing drinking quantity, Subordinate to Others (OR = .76) was marginally related to a lower likelihood of being included in the structural zeros while Silencing Self (OR = 1.48) was significantly related to a greater likelihood of being included in the structural zeros. Finally, Virtuous and Chaste [exp(b) = .80] was inversely related to the count portion of drinking quantity.

Marianismo, PBS, and drunkenness (Table 3)

Table 3.

Marianismo, PBS, and drunkenness

Variable	Stopping/Limiting Drinking (R2 = .03)				Manner of Drinking (R2 = .06*)				Drunkenness (R2 = .31***)
	B	SE	LLCI	ULCI	B	SE	LLCI	ULCI	B	SE	LLCI	ULCI
Virtuous and Chaste	.03	.09	−.07	.30	.17*	.08	.01	.34	−.41*	.19	−.78	−.03
Family Pillar	−.04	.09	−.22	.14	−.11	.08	−.27	.05	.13	.18	−.21	.48
Spiritual Pillar	−.07	.08	−.22	.09	.01	.07	−.13	.14	.04	.16	−.27	.35
Subordinate to Others	.19†	.10	−.01	.22	.05	.09	.13	.23	.41†	.21	−.00	.81
Silencing Self	−.07	.12	−.30	.16	−.05	.11	−.26	.16	.07	.23	−.38	.51
Stopping/Limiting Drinking	−	-	-	-	-	-	-	-	.03	.16	−.29	.35
Manner of Drinking	-	-	-	-	-	-	-	-	−.96***	.20	-1.36	−.56

Note. Loglikelihood: −769.53, AIC: 1589.05, BIC: 1672.96; LLCI = 95% lower limit confidence interval. ULCI = 95% upper limit confidence interval.

^***p< .001.

^**p< .01,

^*p< .05,

^†p< .10.

No marianismo dimensions were related to Stopping/Limiting Drinking, although there was a positive association between Subordinate to Others and Stopping/Limiting Drinking that approached significance (p = .07). Only Virtuous and Chaste was related to Manner of Drinking (positively). Data further indicated that Virtuous and Chaste and Manner of Drinking were each related to a reduced likelihood of drunkenness. In addition, Subordinate to Others was marginally related to an increased likelihood of drunkenness (p = .05). Since the only significant relation to emerge between marianismo and PBS was for the Virtuous and Chaste and Manner of Drinking dimensions, respectively, we analyzed the indirect effect using only these two variables in relation to drunkenness. We found a significant total effect (B = −.04, SE = .02, p = .03) and total natural indirect effect (B = −.02, SE = .01, p = .02). The pure natural direct effect was not significant (B = −.02, SE = .02, p = .34), suggesting the relation between Virtuous and Chaste and drunkenness was accounted for by Manner of Drinking.

Discussion

Preliminary data indicated that the prevalence of past-30 day drinking in this sample was higher than reported drinking rates among Latina/o college students from a national study.⁵² Further, a higher proportion of current drinkers in this sample reported experienced drunkenness than national samples of Latina adult women over the age of 18.⁵³ Overall, results from the primary analyses supported the notion that alcohol use and related behaviors may be, in part, a manifestation of specific dimensions of marianismo among Latina college women. However, the disparate pattern of findings across dimensions suggests that only one dimension of marianismo (i.e., Virtuous and Chaste) may be protective against alcohol-related outcomes. For one, Virtuous and Chaste was negatively related to typical past month drinking quantities. Researchers have posited that, among men, drinking is an expression of masculinity.^11,12 The converse may be true for Latinas and refraining from or limiting alcohol use may be a way that women express aspects of the marianismo gender role. We note there were no relations between marianismo and frequency of drinking, however. It is possible that that the women in our sample may have attended social events and/or other activities that encouraged drinking, as social motivations for drinking are prevalent in college.³⁸ Perhaps frequency of drinking may be the expression of another role for these women (e.g., college student role), while limiting quantity of consumption may be an expression of purity aspects of marianismo.

Results from the second model indicated that, while endorsing Virtuous and Chaste may inhibit Latinas from experiencing drunkenness, endorsing Subordinate to Others may have the opposite effect, although the latter effect was marginal (p < .10). There are multiple explanations that may account for this finding. Previous research among Latina/os indicates dimensions of marianismo measured with the MBS are related to adverse cognitive emotional factors (e.g., depression, hostility¹⁹) and suggests a link between marianismo characterized by submissiveness and depressive symptoms among Latinas.¹⁸ To the extent that Latinas endorse Subordinate to Others and experience distress, it is possible that drinking alcohol may be a way to cope. ³⁹ Another possible explanation is that college women enacting this dimension of marianismo may be more susceptible to drinking-related peer influence. That is, they may be more likely to conform to the actual or perceived behavior of their peers. Research shows that college students may overestimate peer drinking-related norms,⁴⁰ and that perceived norms are strongly related to college students’ own drinking behavior.⁴¹

Aligned with current research, the majority of the sample used PBS,¹ and Manner of Drinking strategies were more strongly and consistently related to decreased alcohol use than Stopping/Limiting Drinking strategies.^23,24 These data suggested Virtuous and Chaste was the only marianismo dimension related to Manner of Drinking and no dimensions were related to Stopping/Limiting Drinking. An inspection of the items in each PBS subscales indicates nuances that may have implications when considered in the context of gender performance. Although both strategies are implemented to avoid engaging in problematic alcohol use, the behaviors associated with Stopping/Limiting Drinking may not be as easily perceived by others in a social environment. However, engaging in drinking games or ‘chugging’ alcohol and the avoidance of such behaviors (Manner of Drinking strategies) may be more obvious behavior in a social context and therefore more obviously consistent with the Virtuous and Chaste aspects of marianismo than the less obvious Stopping/Limiting Drinking strategies of failing to drink more water or having a drink with extra ice. Importantly, the indirect effects model indicated the relation between Virtuous and Chaste and drunkenness was fully accounted for by Manner of Drinking, supporting an indirect associative nature between aspects of Latina femininity and high-risk alcohol use among drinkers. These indirect findings map onto previous qualitative research. For example, scholars describe a seemingly contradictory relation between femininity and drunkenness.²⁹ It is suggested that high levels of intoxication among young women is understood to be somewhat normative behavior but is at the same time considered unfeminine and “trashy”.²⁹ While drunkenness may provide an outlet to express femininity in a way that deviates from what is deemed socially acceptable,^29,42 research suggests women engage in behaviors to inhibit their level of drunkenness from being perceived negatively by others.²⁹ Other research indicates women perceive that intoxication may put them more at risk for being sexually victimized or experiencing other types of unwanted sexual experiences.^42,43 Because of the highly gendered nature of this specific type of risk, engaging in protective practices to minimize such risk could be considered an expression of gender.⁴³ Although our study is the first to examine marianismo and its link to PBS and drunkenness, our findings regarding Virtuous and Chaste are aligned with theoretical perspectives of femininity, drunkenness, and safety discussed in qualitative research.

Limitations and future directions

Among the limitations present in this work, data are cross-sectional and cannot support temporal or causal conclusions. Also, we only measured drunkenness as our high-risk alcohol use variable, although previous research suggests college women may ‘downplay’ the language they use when describing drunkenness (e.g., using words like ‘tipsy’ or ‘buzzed’, even when highly intoxicated), in order to portray themselves as more aligned with the social norms regarding feminine drinking,⁴⁴ rendering the under-reporting of this behavior quite possible. Furthermore, research shows that the gendered patterns of drinking among Latina/os vary as a function of country of origin.⁴⁵Although this sample was predominantly Mexican American, future research should examine the influence of country of origin. Relatedly, research suggests overlap across racial and ethnic groups regarding their respective feminine gender roles (e.g., deference among Vietnamese American women⁴⁶), highlighting the utility of examining if aspects of marianismo that are inversely related to drinking are unique to Latinas, or generalizable to women from other racial and ethnic groups.

A strength of the current study was examining each dimension of marianismo separately to test the unique relation of each dimension to alcohol use; however, this does not reflect the bi-cultural experience of Latinas in the U.S. A high percentage of Latina college students are first-generation students.⁴⁸ and may also experience negotiating between heritage cultural socialization from their home life and U.S. college cultural socialization. Discussions in qualitative research^49,50 suggest that one strategy Latina college students may use is to adopt a dual identity, and enact specific identities depending on the context. Thus, Latinas may choose to enact gender role identities more consistent with U.S. college culture when in the college environment and enact separate gender role identities more consistent with their respective heritage cultures when around family. The concept of a dual identity, as it relates to gender roles and drinking behavior among Latinas, is understudied in the current literature and would greatly enrich future research.

In addition, future research would benefit from expanding upon the current findings with a model including motivations for drinking and college drinking climate. Studies suggest individuals are socially incentivized to drink.⁴⁷ Therefore, if a Latina college student endorses social motivations for drinking and attends a campus with a particularly high prevalence of drinking, these combined factors could (a) reduce or eliminate the extent that marianismo is protective against alcohol use or (b) foster the use of PBS to protect against drunkenness or other adverse consequences related to heavy drinking.

In closing, this study sheds light on the importance of considering the cultural and gender role prescribed mechanisms influencing engagement in drinking behaviors. While this study is the first to examine how certain aspects of marianismo are directly and indirectly associated with alcohol use among Latina college students, there is considerable work left to be done. As such, we encourage scholars to build on the present work so that researchers and practitioners can gain a more nuanced understanding of the ways in which traditional feminine gender roles might influence drinking behaviors among Latina college students.