A cross-sectional examination of race, gender, and intersectionality on protective behavioral strategies and alcohol outcomes among Black and White college students

Abstract

College students consistently report problematic alcohol use (e.g., excessive drinking). Protective behavioral strategies (PBS) are behaviors for reducing alcohol use and/or problems and are negatively associated with alcohol misuse among college students. Studies suggest PBS are used more often and/or more effectively by female individuals. Additionally, examinations specifically between Black and White students have yielded mixed findings and are typically underpowered. Scant studies have investigated the intersectionality between race and gender or examined perceived effectiveness of PBS use related to alcohol outcomes. The current cross-sectional study examined if (1) race, (2) gender, and (3) their interaction moderate the associations between perceived effectiveness of PBS use on PBS use (a path) and between PBS use on alcohol outcomes (i.e., alcohol quantity, peak drinks, and problems; b path). A sample size of N=528 college drinkers ages 18–24 years old (Mage=19.85; 52.5%, Black, 40.3% White) completed an online survey and were used for analyses. A series of moderated mediation models were conducted. Race did not moderate the a or b paths. The positive association for the a path as well as the indirect effect of PBS perceived effectiveness on lower alcohol outcomes through PBS use was stronger for female students than male students. The intersectionality examination indicated a stronger association between more PBS use and lower consumption specifically for White male students and lower problems for Black female students. More attention is needed towards considering intersectionality when identifying at-risk groups and tailoring alcohol intervention efforts.

College drinking and related problems are a public health concern across the US. College students report frequent, risky alcohol use (e.g., one-third of students report binge drinking) at high rates compared to the general population (Hingson, 2017; Patrick et al., 2020; SAMHSA, 2019). Subsequently, students report high levels of alcohol-related problems (Glenn et al., 2022; Hingson, 2017; Patrick et al., 2020), such as having trouble limiting their drinking (>50%) and blacking out from drinking at least once (80%; Glenn et al., 2022). Studies show college students are significantly more likely than young adults not attending college to report negative consequences from drinking (Patrick et al., 2020). It is evident that understanding approaches to reduce risky drinking requires more attention, particularly for college students.

Alcohol protective behavioral strategies (PBS) are behaviors intended to reduce alcohol use and reduce the risk of experiencing negative consequences from drinking (Martens et al., 2007). PBS use has consistently been negatively associated with reports of fewer alcohol problems among college students (see Pearson, 2013 for review). Scant research has investigated if the perceived effectiveness of PBS use (i.e., believing PBS use will reduce negative consequences) is associated with alcohol outcomes and whether this relationship differs across identities. There are known group differences in college drinking across race and gender (Gardner et al., 2020; LaBrie et al., 2011). However, existing research is often limited by a lack of diversity in samples of college students, and few studies have examined intersectionality between race and gender. There is a need to examine if there are racial or gender differences in how PBS use and perceived effectiveness of PBS are associated with alcohol outcomes.

Racial Differences in PBS Use and Drinking

Literature typically suggests White-identifying students report greater problematic drinking behavior (e.g., greater alcohol consumption) compared to students identifying with historically marginalized or minoritized racial/ethnic groups (Banks & Zapolski, 2017; Gardner et al., 2020; Madson & Zeigler-Hill, 2013; Taylor et al., 2022; Zamboanga et al., 2022). However, some research indicates Black individuals may be at greater risk for alcohol-related consequences than White individuals due to differing personal life experiences (e.g., culture, racial discrimination, minority stress; Desalu et al., 2019; Mulia et al., 2009; Vaeth et al., 2017; Witbrodt et al., 2014).

Few studies have examined racial differences in PBS use on alcohol outcomes specifically focusing on Black and White college students. For instance, Black students were not included in two studies examining racial differences in PBS use among college students (Kenney & LaBrie 2013; LaBrie et al., 2011). Some studies did not find racial differences in college student PBS use, but only 2–30% of their samples identified as Black (Clarke et al., 2016; Longo et al., 2018). Other cross-sectional studies with slightly larger percentage of Black participants (32–33%) found support for racial differences, such as certain types of PBS use being related to reduced alcohol consumption (Gardner et al., 2020) and problematic alcohol use (Madson & Zeigler-Hill; 2013) among White but not Black students.

There are no known studies to date examining racial differences in PBS use with adequate representation for Black students relative to White students. Additional research is required with a more representative sample of Black students to understand how identifying as Black versus White may play a role in the relationship between PBS use and alcohol behaviors.

Gender Differences in Drinking and PBS Use

There was a recent update in the guidance from the National Institute of Health (NIH) regarding the definitions of sex (i.e., biological category) and gender (i.e., socially constructed role; NIH, 2022). These terms have been used and defined inconsistently in past research, such as reporting gender but using labels for sex (e.g., male/female) instead. Language in the current examination matches the terms used in the original articles (many dated before the updated guidance) to reflect the responses participants endorsed.

Prior research suggests male students (LaBrie et al., 2011) and men (Madson & Zeigler-Hill, 2013) report higher levels of alcohol consumption than female students and women. Yet, more current research found gender differences in select college drinking behaviors is decreasing (White, 2020). Recently, Schulenberg et al. (2021) found college women reported slightly higher prevalence for past 30-day alcohol use (57% vs. 53%) and being drunk (30% vs. 26%), but more college men reported higher-risk drinking behaviors (e.g., binge drinking; 32% vs. 20%).

Previous research has established differences in PBS use across gender, with female students (Clarke et al., 2016; Kenney & LaBrie, 2013; LaBrie et al., 2011; Longo et al., 2018) reporting more PBS use than male students. Interestingly, Li et al. (2022) found no gender differences when examining the moderating effect of alcohol use levels (low, average, high) between PBS and alcohol consequences. More research is needed to clarify the effect of gender on PBS use and drinking behaviors.

Intersectionality

Although many studies have examined racial or gender differences in college drinking behaviors, research is lacking regarding the intersection between race and gender, which may be associated with differing alcohol behaviors. Examining intersectionality has revealed important distinctions in studies investigating substance use (Holm et al., 2022; Witbrodt et al., 2014). For example, Witbrodt et al. (2014) was able to identify a significantly stronger relationship between heavy drinking and alcohol problems among White men than Black or Hispanic men, yet Black women demonstrated greater odds of alcohol dependence than White women. These findings are in contrast with those of Clarke et al. (2013), where a three-way interaction between race, gender, and alcohol use on alcohol problems was not significant among college students.

Research is more limited when examining intersectionality of PBS use on alcohol outcomes between White and Black college students. Some studies examining race and gender differences in PBS use either do not further examine intersectionality (LaBrie et al., 2011; Lawrence et al., 2010) or have predominantly White samples (Clarke et al., 2016; Demant et al., 2018; Kenney & LaBrie, 2013). For instance, one study that did examine the intersection of race (2% identifying as Black) and gender on the association between PBS use and alcohol problems with college students did not find a significant three-way interaction of the predictors on alcohol problems (Clarke et al., 2016).

Given that documented differences exist across identities (Rogers & Kelly, 2011), it is critical to examine intersectionality for health behaviors particularly among college students, who are at greater risk of problematic alcohol use. It is possible having a small sample of Black students (2–3%) may have contributed to the non-significant findings in existing intersectionality alcohol research (e.g., Clarke et al., 2013; 2016). Exploring the association between PBS use, alcohol outcomes, and intersectionality of race and gender using a sample adequately representative of both Black and White students builds important formative work to address potential disparities in alcohol-related behaviors.

Perceived Effectiveness of PBS use

Research suggests that students who believe PBS are effective for limiting drinking or reducing alcohol problems are more likely to engage in PBS than those who do not believe they are effective (González-Ponce et al., 2022; Pearson, 2013). Prior studies (Bolts et al., 2022; De Leon et al., 2022; Fairlie et al., 2021; Ray et al., 2009; Werch, 1990) have found a positive association between the perceived effectiveness of PBS and the frequency of PBS use among college students. Findings from De Leon et al. (2022) indicate greater levels of perceived effectiveness may be indirectly associated with less drinking and fewer alcohol problems via greater usage of PBS. Bolts et al. (2022) identified that latent profiles with greater levels of PBS perceived effectiveness and use also reported fewer alcohol problems and lower levels of alcohol use. Together, these findings suggest higher PBS perceived effectiveness and PBS use frequency may lead to lower alcohol consumption levels and fewer associated problems. However, the few studies that have examined PBS perceived effectiveness among college students included predominantly White-identifying samples. Research involving racially representative college student samples is needed to ascertain the extent to which differences in PBS perceived effectiveness may correspond with differences in PBS use and college alcohol behaviors.

Current Study

Due to scant studies examining how race and gender may be related to PBS use and alcohol outcomes among college students, and virtually no studies using adequate samples of Black students, further research is required to clarify how race may play a role in how perceived effectiveness of PBS use may impact alcohol outcomes. Therefore, this cross-sectional study aims to examine if (1) race and (2) gender distinctly moderate the sequential pathway from perceived effectiveness of PBS use to PBS use to alcohol outcomes (i.e., drinking quantity, drinking peak, alcohol-related problems) using a sample of college students with a larger proportion of Black students compared to previous studies. Further, given the existing literature demonstrating alcohol-related behaviors may differ across the intersectionality of race and gender, this study also aims to explore (3) if intersectionality moderates the same associations from aims 1 and 2 (see Figure 1).

Figure 1.

Moderated Mediation Model of Perceived Effectiveness of PBS Use, PBS Use, and Alcohol Consumption with Race, Gender, and Their Interaction as a Moderator

Note. PBS = Protective Behavioral Strategy. Race was coded such that 0 = White, 1 = Black. Gender was coded such that 0 = female, 1 = male. N = 521 for the models examining gender as the moderator. N = 394 for the models examining race as a moderator. N = 393 for the models examining intersectionality as a moderator.

Method

Participants were undergraduate students recruited for a longitudinal randomized clinical trial that was designed to examine the effectiveness of emailed post-intervention boosters, which contained either PBS and personalized normative feedback (PNF) or PNF alone, on alcohol use, alcohol-related problems, PBS, and descriptive norms (see Braitman et al., 2021 for details). The current study is a secondary data analysis of the baseline survey data, provided before participants completed the intervention or received booster messages.

Participants and Procedure

Students (N=528) were recruited from a large public university via emailed student announcements and the psychology research pool. To be eligible, students had to be between the ages of 18–24 and have consumed at least one alcoholic beverage in the past two weeks. After providing informed consent, students completed an online survey. A majority of the sample identified as Black (52.5%, n=277) or White (40.3%, n=213). On average, students were 19.85 years old (SD = 1.65) with 71.6% identifying as female. See Table 1 for sample demographics. The study was approved by the university’s institutional review board.

Table 1.

Participant Demographic Information

Variable	M (SD)
Age	19.85 (1.65)
	n (%)
Student status
Freshman	177 (33.5)
Sophomore	150 (28.4)
Junior	96 (18.2)
Senior	102 (19.3)
Other	3 (.6)
Gender
Female	378 (71.6)
Male	149 (28.2)
Ethnicity – Hispanic or Latinx
Yes	58 (11)
No	467 (88.4)
Race
Black	277 (52.5)
Asian	51 (9.7)
Native American	19 (3.6)
White	213 (40.3)
More than one race	63 (12.1)
Only Black or White
Only Black	233 (44.1)
Only White	166 (31.4)

Note. Percentages are of the total sample (N = 528). Percentages for race do not sum to 100% because participants were allowed to select all that apply. There was one missing person for gender (n = 527). Only Black or White include participants who only identified as Black or White and not as multiracial were used in analyses examining race (n = 394).

Materials

Demographic information

Students were asked to provide their age, gender (response options included male, female, trans*, other), and race (response options included Black, Asian or Pacific Islander, White, Native American, other, and participants could select all that apply). Note that although we asked about the social construct of gender, the response options available to participants did not map onto the guidance suggested by NIH in 2020. We present our findings for “male students” and “female students” throughout since these are the identity labels participants selected.

Alcohol Consumption

The Daily Drinking Questionnaire (DDQ; Collins et al., 1985) was used to assess alcohol consumption. Students were presented with a grid of the days of the week and asked how many standard drinks they consumed on each day and over how many hours they consumed them for a typical week in the past 30 days. Alcohol quantity was calculated by summing the number of drinks listed across days. Peak drinks reflected the greatest number of drinks consumed on a single day.

Alcohol-related Problems

The Young Adult Alcohol Consequences Questionnaire (YAACQ; Read et al., 2006) was used to assess alcohol-related problems. Students were presented a list of 48 alcohol-related problems and selected “yes” or “no” to indicate whether they experienced each problem in the past 30 days. The number of problems experienced were summed to create a total score.

PBS Use

The Strategy Questionnaire (SQ; Sugarman & Carey, 2007) was used to assess the frequency of using drinking control strategies in the past 30 days. Students were presented a list of 21 strategies and asked how often they had used them. The response options were altered to reflect an exact number of times used from “never” to “more than 10 times” (Braitman et al., 2015). There are three subscales: selective avoidance, strategies while drinking, and alternatives to drinking. The selective avoidance and strategies while drinking subscales were adjusted by dividing the two subscale scores and by drinking frequency to control for opportunities to use those strategies. This adjustment for drinking frequency prevents curvilinear associations due to the conflation of PBS use with opportunity for its use (drinking occasions) and is appropriate with an absolute frequency response scale (see Braitman et al., 2015 for details). This was not done for the alternatives to drinking subscale since it is not contingent on drinking frequency. To create a total score, the values for the three subscales (including the two adjusted subscales) were summed. Previous research found that both adjusted subscales are negatively associated with alcohol use suggesting that using a total composite score is appropriate (Braitman et al., 2015). Internal consistency was α = 0.93 for the current sample for the overall measure.

Perceived Effectiveness of PBS

A modified version of the SQ (Sugarman & Carey, 2007) was used to assess the perceived effectiveness of drinking control strategies participants used. Each of the original items were repeated, with different instructions asking participants “In the past 30 days, when you used each strategy, how effective was it in helping you to reduce your alcohol use and limit alcohol related consequences?”. Participants rated the perceived effectiveness for each strategy on a 5-point scale from “not effective” to “very effective”. Participants could also endorse they did not use that strategy. The same three subscales are calculated by taking the mean of items/strategies used, with a total score created by taking the mean across all three subscales. Internal consistency was α = 0.95 with the current sample.

Analysis Approach

To examine all three study aims, a series of six moderated mediation analyses were conducted in a structural equation modeling framework in Mplus (version 8.7; Muthén & Muthén, 1998–2021) using maximum likelihood estimation. For each aim, two separate models were conducted to assess alcohol outcomes (i.e., alcohol quantity and problems in one model representing typical drinking [models 1–3], and alcohol peak and problems in a separate model representing highest risk drinking [models 4–6]), resulting in six moderated mediation models. For all models (see Figure 1 for a conceptual model), the a paths reflected perceived effectiveness of PBS use predicting PBS use and the b paths reflected PBS use predicting alcohol outcomes (consumption directly, and problems directly, controlling for consumption). Moderators were examined for both the a and b paths in each model using a product of coefficients approach to examine all aims.

For aim 1 examining race as a moderator (for typical drinking [model 1] and highest risk drinking [model 4]), participants identifying as Black only or White only were included. A two-way interaction term between race and perceived effectiveness of PBS use (a path) or PBS use (b path) was created to examine race as a moderator. Given robust associations with the outcomes of alcohol use and problems in the literature, gender (Wilsnack et al., 2000) was controlled for in aim 1 models. For aim 2 examining gender as a moderator (for typical drinking [model 2] and highest risk drinking [model 5]), the full sample was used. A two-way interaction term between gender and perceived effectiveness of PBS use (a path) or PBS use (b path) was created to examine gender as a moderator. For aim 3 examining the intersection of race and gender as a moderator (for typical drinking [model 3] and highest risk drinking [model 6]), participants identifying as Black only or White only were included. A three-way interaction term between race, gender, and perceived effectiveness of PBS (a path) or PBS use (b path) was included to explore the effect of intersectionality.

Significant moderation results were followed up with simple effects (i.e., the effect for each race [aim 1], each gender [aim 2], or each combination of race and gender [aim 3]). For models without significant interactions, interpretations focused on main effects instead. Significance of indirect effects was determined using bias-corrected bootstrapped 95% confidence intervals with 10,000 replications. Race was coded as 0 = White, 1 = Black. Gender was coded as 0 = female, and 1 = male. Since this study is a secondary examination, this study was not preregistered; however, the original study (Braitman et al., 2022) was preregistered (ClinicalTrials.gov NCT03440463). Data and analysis code for this study are available upon request by emailing the corresponding author.

Results

Normality was confirmed for peak drinks; however, both quantity and alcohol-related problems were positively skewed and a natural log transformation was used to adjust for normality. Outliers (identified via boxplots) were winsorized for peak drinks; none were identified for quantity or problems. See Table 2 for descriptive statistics on perceived effectiveness of PBS, PBS use, and alcohol outcomes across race and gender identities.

Table 2.

Means and Standard Deviations for Perceived Effectiveness of Protective Behavioral Strategies (PBS), PBS use, and Alcohol Outcomes by Race and Gender for Only Black and White Participants

Variable	Gender
	Female	Male
	Race
	Black (n = 171)	White (n = 107)	Black (n = 56)	White (n = 59)
	M (SD)	M (SD)	M (SD)	M (SD)
Perceived effectiveness of PBS use	4.06a (0.83)	4.34ab (.60)	4.09 (0.72)	3.87b (0.79)
PBS use	60.99abc (47.88)	54.32a (37.61)	52.04b (35.66)	45.82c (28.16)
Quantity	5.70ab (6.59)	7.82a (7.57)	8.85bd (10.03)	13.79ad (14.28)
Peak	2.82ab (2.10)	3.93b (2.96)	4.70a (4.47)	6.36b (4.42)
Problems	4.74ad (5.23)	7.93ac (8.52)	5.14bc (6.19)	8.08bd (7.34)

Note: M = Mean; SD = Standard Deviation; PBS = Protective behavioral strategies. Means and standard deviations are from Black and White only sample excluding one person for missing gender (n = 393). Means with matching subscripts across each variable differ at the p = .05 level by independent samples t-tests.

Aim 1: Race as a Moderator

As seen in Table 3, race did not significantly moderate the a path (perceived effectiveness of PBS use predicting PBS use) or b path (PBS use predicting quantity/problems) in the model focused on alcohol quantity (model 1); however, there were a few significant main effects. Identifying as Black (B = 8.52) and greater perceived effectiveness (B = 11.77) were significantly linked to greater PBS use. Identifying as Black was also significantly linked to lower levels of alcohol quantity (B = −0.26) and related problems (B = −0.19), indicating students identifying as Black typically consumed 0.23 fewer standard drinks and reported 0.17 fewer problems than White students, even after controlling for gender and PBS use (both outcomes) and quantity (only problems), when outcome values are exponentiated. Finally, greater PBS use was significantly linked to lower quantity but not to problems when controlling for quantity.

Table 3.

Protective Behavioral Strategy Use as a Mediator Between Perceived Effectiveness and Typical Drinking Outcomes: Interactions with Gender, Race, and Their Interaction

	B	p	95% CI	β
Aim 1: Race Moderation (model 1)
a paths (predicting PBS use)
Perceived Effectiveness → PBS use	11.77	.005	4.413, 20.854	0.216
Race → PBS use	8.52	.026	1.043, 15.980	0.103
Race * Perceived Effectiveness → PBS use	6.60	.199	−3.861, 16.291	0.098
b paths (predicting quantity and problems)
PBS use → Quantity	−0.003	.055	−0.006, 0.000	−0.150
Race → Quantity	−0.26	.001	−0.4102, −0.101	−0.151
Race * PBS use → Quantity	0.002	.274	−0.002, 0.006	0.084
Perceived Effectiveness → Quantity	−0.33	<.001	−0.433, −0.277	−0.295
PBS use → Problems	−0.001	.711	−0.005, 0.004	−0.035
Race → Problems	−0.19	.038	−0.376, −0.009	−0.095
Race * PBS use → Problems	−0.001	.700	−0.006, 0.004	−0.034
Perceived Effectiveness → Problems	−0.04	.589	−0.174, 0.099	−0.028
Aim 2: Gender Moderation (model 2)
a paths (predicting PBS use)
Perceived Effectiveness → PBS use	22.77	<.001	17.78, 28.13	0.413
Gender → PBS use	−12.00	<.001	−18.64, −5.45	−0.131
Gender * Perceived Effectiveness → PBS use	−15.90	<.001	−24.04, −7.67	−0.158
SIMPLE: Male Perceived Eff. → PBS use	6.87	.032	0.742, 13.276
b paths (predicting quantity and problems)
PBS use → Quantity	−0.002	.018	−0.004, 0.000	−0.097
Gender → Quantity	0.39	<.001	0.231, 0.542	0.207
Gender * PBS use → Quantity	0.001	.613	−0.002, 0.004	0.017
Perceived Effectiveness → Quantity	−0.25	<.001	−0.348, −0.153	−0.223
PBS use → Problems	−0.003	.012	−0.005, 0.000	−0.107
Gender → Problems	−0.30	<.001	−0.481, 0.122	−0.136
Gender * PBS use → Problems	0.001	.613	−0.002, 0.004	0.607
Perceived Effectiveness → Problems	−0.08	.171	−0.196, 0.038	−0.167
Simple indirect effects			95% BCCI
Female Perceived Eff. → PBS use → Quantity	−0.05	.026	−0.088, −0.008
Female Perceived Eff. → PBS use → Problems	−0.06	.016	−0.111, −0.012
Male Perceived Eff. → PBS use → Quantity	−0.01	.132	−0.039, −0.001
Male Perceived Eff. → PBS use → Problems	−0.02	.109	−0.050, −0.002
Aim 3: Intersectionality of Race and Gender (model 3)			95% CI
a paths (predicting PBS use)
Perceived Effectiveness → PBS use	15.60	.017	4.204, 29.767	0.292
Race → PBS use	8.59	.063	−0.400, 17.759	0.105
Gender → PBS use	−5.96	.237	−15.744, 3.819	−0.067
Race * Gender → PBS use	−2.74	.719	−17.320, 12.666	−0.024
Race * Perceived Eff. → PBS use	5.86	.423	−10.136, 19.105	0.089
Gender * Perceived Eff. → PBS use	−9.25	.247	−25.929, 5.347	−0.095
Race * Gender * Perceived Eff. → PBS use	−8.53	.427	−28.828, 13.521	−0.057
b paths (predicting quantity and problems)
PBS use → Quantity	−0.002	.411	−0.005, 0.003	−0.074
Perceived Effectiveness → Quantity	−0.32	<.001	−0.428, −0.218	−0.288
Race → Quantity	−0.25	.009	−0.434, −0.056	−0.146
Gender → Quantity	0.34	.005	0.104, 0.582	0.184
Race * Gender → Quantity	0.02	.903	−0.316, 0.356	−0.009
Race * PBS use → Quantity	0.000	.953	−0.0065, 0.004	−0.005
Gender * PBS use → Quantity	−0.01	.099	−0.013, 0.002	−0.134
Race * Gender * PBS use → Quantity	0.01	.062	−0.001, 0.018	0.148
SIMPLE: Black female PBS use → Quantity	−0.002	.155	−0.004, 0.001
SIMPLE: White male PBS use→ Quantity	−0.01	.022	−0.014, −0.001
SIMPLE: Black male PBS use → Quantity	0.01	.141	−0.005, 0.023
PBS use → Problems	0.000	.897	−0.005, 0.005	−0.014
Race → Problems	−0.19	.075	−0.395, 0.016	−0.093
Gender → Problems	−0.22	.178	−0.524, 0.117	−0.100
Race * Gender → Problems	−0.003	.989	−0.427, 0.411	−0.001
Race * PBS use → Problems	−0.003	.368	−0.009, 0.003	−0.088
Gender * PBS use → Problems	−0.002	.672	−0.013, 0.009	−0.041
Race * Gender * PBS use → Problems	0.01	.662	−0.008, 0.019	0.079
Simple indirect effects			95% BCCI
White female Perceived Eff. → PBS use → Quantity	−0.02	.465	−0.110, 0.026
White female Perceived Eff. → PBS use → Problems	−0.01	.905	−0.107, 0.062
Black female Perceived Eff. → PBS use → Quantity	−0.03	.256	−0.089, 0.006
Black female Perceived Eff. → PBS use → Problems	0.05	.130	−0.135, −0.011
White male Perceived Eff. → PBS use → Quantity	−0.12	.120	−0.343, −0.014
White male Perceived Eff. → PBS use → Problems	−0.04	.624	−0.248, 0.073
Black male Perceived Eff. → PBS use → Quantity	0.17	.224	−0.035, 0.530
Black male Perceived Eff. → PBS use → Problems	0.10	.540	−0.144, 0.444

Note. PBS = Protective Behavioral Strategy, Eff. = Effectiveness. Race was coded such that 0 = White, 1 = Black. Gender was coded such that 0 = female, 1 = male. N = 521 for the models examining gender as the moderator. N = 394 for the models examining race as a moderator and N = 393 for the models examining intersectionality as a moderator. Values in bold indicate significant associations. Estimates for quantity and problems were log transformed.

In the model focused on peak drinks (model 4), race was similarly not found to moderate either the a or b paths (see Table 4). There were significant main effects indicating identifying as Black was related to significantly fewer peak drinks (B = −1.18, p <.001) and greater PBS use (p = .026).

Table 4.

Protective Behavioral Strategy Use as a Mediator Between Perceived Effectiveness and Peak Drinking Outcomes: Interactions with Gender, Race, and Their Interaction

	B	p	95% CI	β
Aim 1: Race Moderation (model 4)
a paths (predicting PBS use)
Perceived Effectiveness → PBS use	11.77	.005	4.413, 20.855	0.216
Race → PBS use	8.52	.026	1.044, 15.980	0.103
Race * Perceived Effectiveness → PBS use	6.60	.199	−3.862, 16.291	0.098
b paths (predicting drinking and problems)
PBS use → Peak	−0.01	.253	−0.018, 0.003	−0.093
Race → Peak	−1.18	<.001	−1.733, −0.625	−0.195
Race * PBS use → Peak	0.01	.179	−0.005, 0.023	0.107
Perceived Effectiveness → Peak	−1.03	<.001	−1.405, −0.671	−0.260
PBS use → Problems	−0.002	.440	−0.006, 0.003	−0.072
Race → Problems	−0.18	.065	−0.365, 0.016	−0.088
Race * PBS use → Problems	−0.001	.672	−0.006, 0.004	−0.037
Perceived Effectiveness → Problems	−0.09	.214	−0.233, 0.050	−0.068
Aim 2: Gender Moderation (model 5)
a paths (predicting PBS use)
Perceived Effectiveness → PBS use	22.77	<.001	17.781, 28.130	0.413
Gender → PBS use	−12.00	<.001	−18.64, −5.444	−0.131
Gender * Perceived Effectiveness → PBS use	−15.90	<.001	−24.04, −7.673	−0.158
SIMPLE: Male Perceived Eff. → PBS use	6.87	.032	0.742, 13.276
b paths (predicting peak and problems)
PBS use → Peak	−0.002	.484	−0.007, −0.004	−0.027
Gender → Peak	1.76	<.001	1.164, 2.374	0.271
Gender * PBS use → Peak	0.002	.481	−0.003, 0.007	0.010
Perceived Effectiveness → Peak	−0.76	.001	−1.112, −0.400	−0.195
PBS use → Problems	−0.004	.001	−0.006, −0.001	−0.149
Gender → Problems	−0.33	<.001	−0.523, −0.146	−0.148
Gender * PBS use → Problems	0.002	.481	−0.003, 0.007	0.029
Perceived Effectiveness → Problems	−0.12	.057	−0.238, 0.003	−0.087
Simple indirect effects
Female Perceived Eff. → PBS use → Peak	−0.04	.488	−0.163, 0.080
Female Perceived Eff. → PBS use → Problems	−0.08	.002	−0.139, −0.033
Male Perceived Eff. → PBS use → Peak	−0.01	.539	−0.066, 0.021
Male Perceived Eff. → PBS use → Problems	−0.03	.081	−0.063, −0.004
Aim 3: Intersectionality Moderation (model 6)			95% CI
a paths (predicting PBS use)
Perceived Effectiveness → PBS use	15.60	.017	4.204, 29.767	0.292
Race → PBS use	8.59	.063	−0.400, 17.759	0.105
Gender → PBS use	−5.96	.237	−15.744, 3.819	−0.067
Race * Gender → PBS use	−2.74	.719	−17.320, 12.666	−0.024
Race * Perceived Eff. → PBS use	5.86	.423	−10.137, 19.105	0.089
Gender * Perceived Eff. → PBS use	−9.25	.247	−25.929, 5.347	−0.095
Race * Gender * Perceived Eff. → PBS use	−8.53	.427	−28.828, 13.521	−0.057
b paths (predicting peak and problems)
PBS use → Peak	−0.01	.862	−0.013, 0.013	−0.015
Race → Peak	−1.14	<.001	−1.787, −0.539	−0.188
Gender → Peak	1.67	.001	0.745, 2.653	0.252
Perceived Effectiveness → Peak	−1.01	<.001	−1.365, −0.634	−0.252
Race * Gender → Peak	0.06	.923	−1.225, 1.417	0.008
Race * PBS use → Peak	0.002	.830	−0.014, 0.015	0.017
Gender * PBS use → Peak	−0.02	.148	−0.053, 0.011	−0.137
Race * Gender * PBS use → Peak	0.03	.148	−0.012, 0.073	0.141
PBS use → Problems	−0.001	.660	−0.006, 0.004	−0.045
Race → Problems	−0.17	.117	−0.393, 0.041	−0.06
Gender → Problems	−0.25	.123	−0.558, 0.079	−0.115
Perceived Effectiveness → Problems	−0.08	.255	−0.224, 0.059	−0.062
Race * Gender → Problems	−0.001	.998	−0.429, 0.422	−0.000
Race * PBS use → Problems	−0.003	.310	−0.009, 0.003	−0.098
Gender * PBS use → Problems	−0.003	.604	−0.014, 0.008	−0.050
Race * Gender * PBS use → Problems	0.03	.148	−0.007, 0.020	0.094

Note. PBS = Protective Behavioral Strategy, Eff. = Effectiveness. Race was coded such that 0 = White, 1 = Black. Gender was coded such that 0 = female, 1 = male. N = 521 for the models examining gender as the moderator. N = 394 for the models examining race as a moderator and N = 393 for the models examining intersectionality as a moderator. Values in bold indicate significant associations. Estimates for quantity and problems were log transformed.

Aim 2: Gender as a Moderator

As shown in Table 3, gender was found to be a significant moderator for the a path (B = −15.90, p <.001), but not the b path in the model focused on alcohol quantity (model 2). Endorsing a female gender identity strengthened the association between perceived effectiveness of PBS use and PBS use (B = 22.77) compared to the simple slope for males (B = 6.87), though both associations were significant. Although there was not a significant interaction for the b path, there were a few significant main effects indicating identifying as male was related to greater quantity (B = 0.39) but fewer problems (B = −0.30). Engaging in more PBS use was related to lower quantity (B = −0.002) and fewer problems (B = −0.003). PBS use was found to significantly mediate the association between perceived effectiveness of PBS and all alcohol outcomes across gender (see Table 3). However, among participants identifying as female, there were stronger indirect effects for both quantity (B = −0.05) and problems (B = −0.06) than participants identifying as male (B = −0.01, −0.02, respectively). Female students typically consumed 0.05 standard drinks and experienced 0.06 alcohol problems less through greater PBS use for every one unit increase in perceived effectiveness of PBS use. Male students typically consumed 0.01 fewer drinks and experienced 0.02 fewer problems (when outcome values are exponentiated), controlling for PBS use and perceived effectiveness of PBS use.

In the model focused on peak drinks (model 5), gender again moderated the a path but not b path, with stronger associations for female students. Indirect effects were significant only for perceived effectiveness related to problems for both female (B = −0.08) and male students (B = −0.03), and there were no significant findings for peak drinks. Results indicate perceived effectiveness may be related to experiencing fewer problems from drinking through PBS use more so for female students than for male students. Similar to the model focusing on quantity, there were some significant main effects for the b path indicating identifying as male was associated with greater peak drinks (B = 1.76) but fewer problems (B = −0.33). Greater PBS use was significantly related to experiencing fewer problems (B = −0.004) but not peak drinks (see Table 4).

Aim 3: Intersectionality of Race and Gender

For the model focused on alcohol quantity (model 3), there was not a significant three-way interaction for either the a or b path; however, there was an interaction approaching significance (see Table 3) for the b path between race, gender, and PBS use related to quantity (B = 0.01, p = .062). This was not significant for problems. Participants identifying as a White male (B = −0.01) were found to have a stronger association between more PBS use and lower quantity (simple slopes for b path are included in Table 3) than other identities. Estimate values remain similar when exponentiated. White male students were expected to consume 0.01 fewer drinks as PBS use increased by one unit. The association between PBS use and quantity was not significant for students identifying as a White or Black female or as a Black male, but the association for Black female students approached significance (B = −0.002). Furthermore, simple indirect effects indicated perceived effectiveness of PBS use had a significant indirect effect on quantity through PBS for White male students (based on 95% BCCIs) and a marginally significant effect for Black female students (based on 90% BCCIs), as well as on problems for only Black female students (see Table 3).

For the model focused on peak drinks (model 6), there was not a significant three-way interaction for either the a or b path. No other simple indirect effects were significant (see Table 4).

Discussion

Across three aims, the present study was the first to examine the moderating role of race (i.e., Black vs. White; Aim 1), gender (i.e., female vs. male; Aim 2), and the intersection of race and gender (Aim 3) on the sequential pathway from perceived effectiveness to PBS use to alcohol outcomes with a relatively large sample of Black-identifying students (52.5%).

For aim 1, there was no support found for race as a moderator between perceived effectiveness to PBS use (a path) or between PBS use to alcohol outcomes (b path). However, consistent with previous research conducted at historically Black colleges that has repeatedly demonstrated that Black college students drink less than White college students (e.g., Meilman et al., 1995; Kapner, 2008), Black students in the present sample reported considerably less alcohol use quantity, fewer drinks on peak drinking day, and fewer alcohol-related problems across gender identities according to main effects. We also found that Black students reported greater PBS use than White students, as similarly reported in previous studies (Gardener et al., 2020; Madson & Zeigler-Hill; 2013).

There was partial support found for aim 2 as gender significantly moderated the relationship between perceived effectiveness of PBS and PBS use, but not from PBS use to alcohol outcomes. Moreover, consistent with numerous studies that have shown that females report more PBS use compared to males (i.e., Benton et al., 2004, Delva et al., 2004, LaBrie et al., 2011, Lewis et al., 2010, Palmer et al., 2010, Walters et al., 2007), in the present sample we found that female students reported higher rates of PBS use than male students.

Regarding the intersection of gender with race for aim 3, Black male students reported higher perceived effectiveness of PBS use than White male students, but the inverse was true for female students according to descriptive findings. These findings regarding PBS use and perceived effectiveness of PBS use have not previously been presented in the literature.

In addition to the descriptive findings, results from our path analyses pertaining to aim 3 revealed interesting findings related to intersectional identities. Specifically, intersectional identity had marginal significance in moderating the b path (i.e., the link between PBS use and alcohol outcomes). We report these findings regarding marginal significance because the examinations focused on intersecting identities parsed the sample into smaller groups, and although the current sample had much better representation of Black students than previous work in this area, it still resulted in fairly small sample sizes (and reduced power) for the intersecting identities, in particular because of the smaller percentage of male students in the sample. The strongest negative associations between PBS and alcohol use quantity were found for only White male and Black female students when looking at the simple slopes. In contrast, the well-established negative association between PBS use and alcohol quantity (cf. Pearson, 2013) was not significant for White female students or Black male students. Additionally, indirect effects revealed perceived effectiveness of PBS use to be most strongly related to less alcohol quantity through PBS use for only White male and Black female students, and to experiencing fewer problems for only Black female students.

Although the relationship between PBS use and alcohol outcomes was found to be significant in many models, and for select individuals, the interaction did not reach significance in the overall model examining race intersecting with gender. This may be because the sample was unbalanced across groups, and although better representation than most studies, there was still a small number of students who identified for specific intersectional identities (e.g., Black male or White male). Taken together, we found that having two minoritized identities or having two majority identities were the most protective intersections of identity, whereas having one minoritized and one majority identity was associated with the least protection.

It is possible that factors that were not measured in the current study may help explain these findings. For example, deviance regulation theory (Blanton & Christie, 2003) indicates that behaviors are based on social identity and social norms. Dvorak and colleagues (2015), have shown that the relationship between social identity and normative beliefs is complicated and that there are situations where individuals may choose risky behaviors that align with their beliefs about the social environment in which they live. Prior work has shown that individuals do hold different norms across different racial and gender identities (Braitman & Lau-Barraco, 2020; Lewis et al., 2009; McCabe et al., 2019). Additional research is needed to understand the social identities and normative beliefs across intersectional identities to better explain the present findings.

This study has a number of strengths including a larger proportion of the sample that was Black-identifying compared to other studies in the field, an examination of the links between perceived effectiveness of PBS, PBS use, and alcohol outcomes by race, gender, and the intersection of race and gender, and the identification of racial and gender identities associated with greater risk and protection from alcohol-related problems. However, findings should be interpreted in light of the limitations. First, this was a cross-sectional survey which precludes causal links and examining the temporal order. Even though we tested the sequential pathway from perceived effectiveness of PBS to PBS use to alcohol outcomes, we cannot rule out that other sequential pathways exist (e.g., that alcohol outcomes lead to PBS use). Second, this research was conducted at a single institution which limits generalizability to the broader population of college students. Third, while the focus of the present study was on differentiating the relationship between White and Black students, we had an underrepresentation of other races and ethnic groups which precluded further comparisons beyond White vs. Black racial identities. As noted above, there was also an unbalanced representation across intersecting identities, with sample sizes for Black and White male students being nearly less than half relative to the female identifying groups. Related, the question we used to assess gender identity was not ideal and both conflated sex and gender terms and did not allow for examination of other gender identities (e.g., non-binary).

Despite the limitations described above, the present study adds to the literature by presenting alcohol use and PBS use patterns across intersections of racial and ethnic identities. Often in the college drinking literature, PBS use is viewed as a predictor of alcohol outcomes, and nuanced predictors of PBS use are typically not examined. The few prior studies that have examined racial and gender identities often used samples without sufficient representation of minority-identifying individuals. Future research is needed to elucidate the underlying mechanisms driving the patterns of responses shown and specifically to better understand the added risk and protection conferred from multiple intersecting minoritized identities vs. the combination of one minority and one majority identity. The clinical implications of the results of the present study are that while identifying as Black and female or White and male may be protective factors, this may be most true for those who hold both of these identities, and special attention should be paid to Black males and White females who may be at greater risk for alcohol problems and may be less inclined to use PBS to reduce their alcohol use.

Public Significance Statement:

This study suggests there to be differences across racial and gender identities in how protective behavioral strategies used to reduce alcohol problems and its perceived effectiveness may be associated with alcohol outcomes. The potential usefulness of these strategies may be more observable for Black female and White male students. It may be important to consider intersectionality between racial and gender identities when identifying at-risk groups and tailoring alcohol intervention efforts.