Examining gender and the longitudinal effect of weight conscious drinking dimensions on body mass index among a college freshman cohort

Abstract

Objective: This study aims to: (1) examine gender differences for weight conscious drinking among college students accounting for the broader phenomenon (e.g. including the Alcohol Effects dimension); and (2) longitudinally examine the effect of weight conscious drinking behaviors on body mass index (BMI). Participants: United States freshmen students from eight participating universities (N= 1,149). Methods: Structural equation modeling was used to model the effect of gender on weight conscious drinking dimensions at 7-month follow-up. Results: Findings suggest a significant effect of gender on Alcohol Effects (β = −.15, SE = .05, p = .005) at 7-month follow-up among college freshmen. Weight conscious drinking dimensions predicted no significant change in BMI at 7-month follow-up among college freshmen. Conclusion: Findings contribute to weight conscious drinking theory and provide campus weight conscious drinking prevention initiatives with evidence to tailor their programming to address female tendencies to engage in compensatory strategies to enhance the psychoactive effects of alcohol.

Keywords:

• College binge drinking
• eating disorders
• weight conscious drinking

Introduction

The attempt to compensate for energy consumed from alcohol, termed “drunkorexia” by some scholars,^1–4 lacks consistent operationalization in the college alcohol literature.⁵ This phenomenon is holistically conceptualized in this paper as “weight conscious drinking” and includes the manifestation of disordered eating behaviors to enhance the psychoactive effects of alcohol, individually or in combination with behaviors aiming to offset the consumption of alcohol-related calories (behaviors can take place before, during, and/or after a drinking episode). Individuals who engage in weight conscious drinking, male or female, may utilize extreme dietary restraint, intensive exercise, and/or purging behaviors (i.e. use of laxatives, diuretics, etc.) in an effort to achieve the desired outcome.

Literature inconsistencies in the operationalization of weight conscious drinking and weight conscious drinking literature findings have stemmed from various factors, including discrepancies across the: (a) scope of the compensatory behaviors manifested across college students; (b) dimensionality of compensatory behaviors; (c) temporality of compensatory behaviors (i.e. before, during, or after drinking episodes); (d) specific intent to compensate for the intake of alcohol-related calories; and (e) target gender.⁶ While some scholars acknowledge the use of exercise as an alcohol-related compensatory behavior,^1,3,7–11 others have restricted the breadth of their empirical examination to dietary restraint.^2,4,12 In addition, literature operationalization discrepancies also exist between investigations that acknowledge the presence of behaviors used to enhance the psychoactive effects of alcohol^8,11,13 alone or in conjunction with alcohol-related compensatory behaviors as part of the weight conscious drinking phenomenon and those that do not.^{2–4,7,9,10,12} In addition to literature discrepancies in the scope of alcohol-related compensatory behaviors and number of latent dimensions present in weight conscious drinking, scholars have also differed in the dimensionality (i.e. severity spectrum from non-clinical to clinical symptomology) measured for alcohol-related compensatory behaviors. For instance, some scholars have highlighted the relevance of accounting for a severity spectrum across alcohol-related compensatory behaviors via the addition of items capturing various levels of exercise^1,9 or dietary restriction.^8,14

Empirical findings suggest that college students are among those most at-risk for weight conscious drinking behaviors,^1–5,15 largely due to a cultural fixation on both body image ideals and heavy drinking behaviors.^11,16 The prevalence of disordered eating among college students has been estimated to be as high as 10–20%, with females being at the higher end of this range, while the prevalence of binge drinking has been reported to be over 40%. Binge drinking increases the risk of numerous health and social problems, such as unplanned sexual activity, alcohol-impaired driving, and violence.^17–19 A literature review of college female alcohol consumption suggested binge drinking was a risk factor for female sexual victimization, with most rapes taking place “when the victim was too intoxicated to resist.”²⁰

Weight conscious drinking among college students predicts binge drinking frequency.⁶ While these behaviors have been individually associated with a variety of negative health outcomes, the potential risks are compounded when disordered eating and binge drinking behaviors occur in tandem. In addition to acute risks related to alcohol toxicity, suicide, motor vehicle accidents, etc., weight conscious drinking behaviors increase an individual’s risk for numerous long-term health consequences, such as hypertension, cardiovascular disease, gastrointestinal cancer, and liver disease.

Previous research suggests that gender may be a significant factor in understanding the weight conscious drinking phenomenon among college students; however, findings regarding gender and weight conscious drinking remain mixed. Social constructionism theory provides a theoretical lens from which to elucidate the gender differences observed in weight conscious drinking. A social constructionist perspective proposes social context shapes the construction of gender roles.²¹ In examining the social structuring of gender within a health behavior context, this theory posits that health-related beliefs and behaviors are by-products of everyday social interactions and institutional structures exerting influence on the extent to which individuals conform to dominant gender norms, commonly stereotyped as femininities or hegemonic masculinities. Strong conformity to dominant gender norms has uncovered differential health risks for males and females alike. Findings from several studies indicate that college females may have a greater tendency to engage in weight conscious drinking behaviors than college males. For example, one study detected a stronger association between weight loss behaviors and alcohol consumption among college females,⁹ and another found a significant increased tendency for college females to resort to self-imposed caloric restriction prior to drinking.⁴ In addition to the tendency to engage in weight conscious drinking behaviors, college males and females may differ with respect to the application and implementation of these behaviors. Barry et al. (2013) reported that males engaged in more physical activity to compensate for calories from alcohol, while females opted for bulimic-like weight management strategies such as skipping meals or self-induced vomiting. Similarly, a study conducted by Gorrell, Walker, Anderson, Boswell²² detected a positive association between bulimic compensatory behaviors and female college undergraduates in their sample. Additionally, another study revealed significant positive effects for females on dietary restriction to enhance the psychoactive effects of alcohol, and two studies revealed significant positive effects for females on compensatory behaviors for the intake of alcohol-related calories.^8,23 Conversely, another study examining the role of gender among male and female college students’ weight conscious drinking¹¹ revealed a significant difference between the percentage of males and females engaging in weight conscious drinking behaviors, with males reporting higher percentages than females across several Compensatory Eating Behaviors in Response to Alcohol Consumption Scale (CEBRACS) items within the Bulimia and Restriction subdomains. An ordinal regression conducted in the same study pointed to higher odds of engaging in weight conscious drinking among masculine-oriented respondents, regardless of biological sex.

Although a number of studies have identified gender-based differences in weight conscious drinking behaviors among college students, the evidence as a whole remains mixed. This lack of agreement between studies may likely be attributed to the varying methods by which weight conscious drinking has been assessed. Several studies have not explicitly related weight management behaviors with the intent to compensate for alcohol-related calories in their assessment of weight conscious drinking. Furthermore, the assessment of weight conscious drinking has largely been through combinations of individual constructs, rather than accounting for the broader phenomenon as a whole (i.e. Alcohol Effects, Dietary Restraint and Exercise, Bulimia, and Restriction). For instance, one study operationalized weight conscious drinking with a single item by asking students to report how many times in the past 30 days they restricted energy intake before drinking alcohol.⁴ Other studies utilizing the CEBRACS to examine gender differences within the broader phenomenon of weight conscious drinking have not applied latent variable analysis to fully account for the multidimensional nature of weight conscious drinking.^11,22 This general lack of standardization, with a comprehensive assessment tool, could conceivably underlie the current discrepancies in the literature.

In addition to the identification of individual risk factors, efforts to prevent weight conscious drinking in the college setting could substantially benefit from research elucidating the effect of these strategies on body weight. However, at present, few studies have investigated the role of weight conscious drinking behaviors on body weight or BMI (BMI). While Barry et al observed negative relationships between weight loss behaviors and exercise on BMI among college males and females, exercise and weight loss behaviors were not measured specifically in response to compensation for the intake of alcohol-related calories. The relationship between behaviors intended to enhance alcohol effects on BMI also remains unaddressed. Moreover, the current literature surrounding alcohol-related compensatory behaviors only includes cross-sectional evaluations. Longitudinal examinations of weight conscious drinking behaviors are needed, to fully understand its consequences and impact on student well-being. Such information will be valuable for tailoring programs to prevent weight conscious drinking behaviors. This information could also potentially dispel myths about the effectiveness of weight conscious drinking compensatory strategies for weight management.

To our knowledge, no studies have specifically assessed the effect of latent weight conscious drinking dimensions on BMI across a college freshman cohort or addressed the effect of gender on each weight conscious drinking dimension. Therefore, the objectives of this study were to examine: (1) the direct effects of weight conscious drinking dimensions on BMI at 7-month follow-up; and (2) the direct effect of gender on weight conscious drinking dimensions among a nonclinical (i.e. not clinically diagnosed with eating or alcohol use disorders) cohort of college freshmen.

Method

Participants and procedure

The 21-item Compensatory Eating Behaviors in Response to Alcohol Consumption Scale (CEBRACS) was administered to a sample of 1,149 freshmen college students at eight U.S. universities during the development phase of the “Get FRUVED ” study. The institutions participating in this project had faculty members of an established a multistate research team. The eight universities are located throughout the southeast, northeast, and Midwest areas of the U.S. Seven of the eight universities are part of the public land-grant system. Student participants completed a battery of surveys along with anthropometric measurements onsite at each participating university, at the beginning of the fall and end of the spring semesters of the 2015–2016 academic year. Participant data regarding health-related behaviors, attitudes, and beliefs were then used to inform future intervention activities to be tested on other college campuses. Details of participant recruitment and enrollment were described previously.²⁴ Briefly, participants were recruited via email messages to first year students and flyers posted on campus and distributed during orientation events at each university. Interested participants completed a brief eligibility screening survey, and those eligible were invited to attend in person assessment visits on their respective campuses. The IRB at the University of Tennessee approved research activities at the University of Tennesse, West Virginia University, South Dakota State University, Syracuse University, and University of Maine. University of Florida, Auburn University, and Kansas State University Institutional Review Boards provided approval for research on their respective campuses. All participants completed an informed consent. This study was registered at Clinicaltrials.gov with the identifier code NCT02941497.

Inclusion criteria

Requirements for eligibility included those specified from the formative phase of the Get FRUVED study, which sought to learn about health behaviors, attitudes, and beliefs of college students at risk for poor health behaviors in order to inform future interventions. Specifically, investigators were focused on diet, physical activity, and stress management behaviors. Eligibility criteria included: freshman student status, 18 years of age or older, and self-reported intake of less than 2 cup equivalents (CE) of fruits and/or less than 3 CE of vegetables per day. Participants were also required to satisfy at least one of the following criteria: (a) BMI ≥ 25 kg/m²; (b) first generation college student; (c) overweight or obese parent(s); (d) economically disadvantaged background; (e) racial minority. Over 80% of students screened were deemed eligible based on these criteria. Participants that indicated consumption of at least one-alcohol containing beverage within the past three months were considered in this analysis of weight conscious drinking behaviors.

Measures

All participants attended an on-site assessment visit on each campus at the time of enrollment and at the end of their first academic year. After providing consent, participants underwent anthropometric measurements, with trained research assistants, and completed survey questions via a secure Web-based platform. Data were obtained on participant sociodemographic characteristics, height and weight to calculate BMI, weight intentions, and weight conscious drinking risk.

Sociodemographic characteristics

Participants’ age, sex assigned at birth (male/female), gender, race/ethnicity, living location (on/off campus), varsity athlete (yes/no), Fraternity/sorority affiliation (yes/no), binge drinking frequency, and university of enrollment were collected within the survey. Gender was measured with the following survey question: “What is your current gender identity?” Gender response options were the following: (1) Male; (2) Female; (3) Trans male/Trans man; (4) Trans female/Trans woman; (5) Gender queer/Gender non-conforming; (6) Different identity; and (7) Choose not to answer. Binge drinking frequency was assessed via the third item from the Alcohol Use Disorders Identification Test-Consumption (AUDIT-C) short form.²⁵ The item stated: “How often did you have six or more drinks on one occasion in the past year?” Binge drinking frequency item response options were the following: (1) Never; (2) Less than monthly; (3) Monthly; (4) Weekly; and (5) Daily or almost daily. As noted in the college alcohol literature, at-risk for binge drinking status was derived from student-reported U.S geographic region of residence. Northeastern and Midwestern U.S. regions were coded as at-risk, and Southeastern, Southwestern, and Northwestern regions were coded as not at-risk.

Body mass index

Trained research assistants measured participants’ height and weight in private research space on each university campus. BMI was calculated as participants’ weight in kilograms divided by their height in meters squared. Calibrated scales were used to weigh participants in kilograms and full-length stadiometers were used for height (meters) at each time point. To ensure greater accuracy, each measure was recorded twice. If the first measure was discrepant from the second by more than one tenth of a decimal, the second measure was repeated.

Weight intentions

Weight intentions were measured at baseline via the following item: “Which of the following are you trying to do about your weight?” Five response options were provided (lose weight, gain weight, stay the same weight, I am not trying to do anything about my weight, and choose not to answer).

Weight conscious drinking

The Compensatory Eating Behaviors in Response to Alcohol Consumption Scale (CEBRACS) was used to measure weight conscious drinking risk. Formerly validated on an undergraduate sample of college males and females, this 21-item self-report measure yielded a coefficient alpha of 0.89 for the total CEBRACS. The following four factor structure from CEBRACS with their corresponding coefficient alphas: (1) Alcohol Effects—behaviors aimed at enhancing the psychoactive effects of alcohol (0.95); (2) Bulimia—bulimic-like behaviors, such as self-induced vomiting (0.92); (3) Dietary Restriction and Exercise (0.87); and (4) Restriction—extreme dietary restriction, such as meal skipping (0.79).¹⁴ Factor 1 contains 7 items that index behaviors designed to enhance the effects of alcohol, and was thus labeled Alcohol Effects; Factor 2 contains 6 items that reflected bulimic-like behaviors (purging) and was named Bulimia; Factor 3 has 6 items that reflected exercising and dietary restraint (Dietary Restraint and Exercise), and Factor 4 has 2 items that indexed extreme Restriction (skipping meals or not eating for a day). The scale is further comprised of three sections accounting for compensatory behavior temporality with respect to drinking episodes (i.e. before, during, or after drinking) and captures behavior frequency on a 5-point Likert scale (1-Never to 5-Almost all the time). For instance, the first item on the scale is worded as follows: “In the past three months, I have eaten less than usual during one or more meals before drinking to get DRUNKER.” The response options for each item on the scale are as follows: (1) Never; (2) Rarely (about 25% of the time), (3) Sometimes (about 50% of the time), (4) Often (about 75% of the time), and (5) Almost all the time. Higher total scores on the scale denote greater risk for weight conscious drinking. Lack of engagement in weight conscious drinking behaviors was determined by those who reported to never engage in any of the weight conscious drinking behaviors described in the CEBRACS.

Statistical analysis

Chi-square tests were used for bivariate data analyses examining associations across participant demographic characteristics, baseline weight intentions, and engagement in weight conscious drinking behaviors. A Bonferroni correction, with an adjusted alpha value of 0.02 (0.05/3), was used to account for multiple post-hoc comparisons. Structural equation modeling (SEM) using a robust weighted least square means and variances method (WLSMV) was used to model hypothesized directional relations between weight conscious drinking dimensions and observed mixed-type variables (i.e. gender and BMI at 7-month follow up). Global model fit was assessed using the following relative fit indices and corresponding cutoffs: Comparative fit index (CFI) and Tucker-Lewis Index (TLI) values greater than .90 and root mean square error of approximation (RMSEA) values less than .06.²⁶ Model estimation was conducted using Mplus version 7.4.²⁷

Missing data

Missing data was over 30% across all CEBRACS items (ranged from 34.9% to 35.4%) and 28.2% for BMI at follow-up. Lower missing values were identified for sex (2.3%), gender (2.5%), and baseline BMI (2.6%). Little’s missing completely at random (MCAR) test confirmed the missing values were not missing completely at random (χ2 [190] = 1,230.610, p ≤ .001). The missing data mechanism was further assessed via the use of missing indicators for each item on the CEBRACS and for BMI at follow-up. Statistical analyses from the available-case analysis (pairwise deletion) were compared for model result triangulation against a missForest single imputation model without missing indicators and a missForest single imputation model containing missing indicators. Results for the missForest single imputation model including CEBRACS and BMI at follow-up missing indicators are provided in Table 4. Imputations were performed in R Version 1.0.153.²⁸

Results

Participant characteristics and model fit

Of the 1,149 students enrolled in the study in the fall of 2015, 860 (74.8%) returned for the second assessment visit in the spring of 2016. Most participants were female sex (64.8%), identified as female gender (64.2%), White (71.1%), non-Hispanic (79.3%), resided on their college campus during their first academic year (83.6%), and intended to lose weight (56.4%). Respondent sex and gender variables were similarly distributed (i.e. approximately 33% male and 64% female, with 3% missing data). A bivariate correlation between sex and gender in the sample yielded a significant strong positive correlation (r = 0.9, p < .001). Given gender minorities comprised less than one percent of the sample and the sex and gender variables were strongly positively correlated, gender was represented in the sample as either male or female. All gender evaluations carried out henceforth in this paper are interpreted with the understanding that the sample under study predominantly identified with either male or female gender, which corresponded to the biological sex they were assigned at birth.

CEBRACS item-level means indicated students reported engaging the most often (i.e. about 25% of the time, on average) in the following two Dietary Restriction and Exercise items: “In the past three months, I have exercised before drinking to make up for the calories in alcohol that I anticipated consuming” (1.5 ± .95) and “In the past three months, I have exercised to make up for the calories in alcohol that I had consumed previously while I was under the effects of alcohol” (1.6 ± 1.03), with the largest variability present across the “After—exercising” item. CEBRACS item-level means for the Bulimia dimension exhibited the lowest averages (ranging from 1.03 to 1.04) and least amount of variability (SD ranging from .20 to .26), with most students reporting “never” engaging in bulimic compensatory behaviors. All four CEBRACS subscales were strongly positively correlated (r ≥ .75) and internal consistency for each dimension was acceptable (α ≥ .63). Participant characteristics, scale correlations, and composite reliabilities for the variables assessed are detailed in Tables 1 and 2, respectively.

Table 1. Participant demographic characteristics and self-reported compensatory behaviors in response to alcohol consumption (n = 1,149).^a

Variables	Frequency n (%)		Missing n (%)
Sex			27 (2.3)
Male	377 (32.8)
Female	745 (64.8)
Gender identity			29 (2.5)
Male	376 (32.7)
Female	738 (64.2)
Other	6 (0.5)
Ethnic origin			34 (2.9)
Hispanic	204 (17.8)
Non-Hispanic	911 (79.3)
Race			0
Black or African American	162 (14.1)
Asian/Asian American	137 (11.9)
White	817 (71.1)
Residence			21 (1.8)
On campus	961 (83.6)
Off campus	167 (14.5)
Athlete Status			36 (3.1)
Athlete	68 (5.9)
Non-athlete	1,045 (90.9)
Greek affiliation			28 (2.4)
Affiliated	11 (1.0)
Nonaffiliated	1,110 (96.6)
Binge drinking frequency			411 (35.8)
Never	378 (32.9)
Less than monthly	230 (20)
Monthly	84 (7.3)
Weekly	44 (3.8)
Daily	2 (0.2)
BDF males			125 (33.2)
Never	94 (24.9)
Less than monthly	82 (21.8)
Monthly	46 (12.2)
Weekly	29 (7.7)
Daily	1 (0.3)
BDF females			265 (35.6)
Never	281 (37.7)
Less than monthly	146 (19.6)
Monthly	37 (5.0)
Weekly	15 (2.0)
Daily	1 (0.1)
Binge drinking region			98 (8.5)
At-risk	528 (46)
Not at-risk	523 (45.5)
Weight intentions			10 (0.9)
Lose weight	648 (56.4)
Gain weight	143 (12.4)
Stay the same weight	208 (18.1)
I am not trying to do anything about my weight	140 (12.2)
	Mean	SD	Missing (%)
Age	19.20	0.86	0
BMI (kg/m^2) baseline	24.17	4.32	2.6
BMI (kg/m^2) follow-up	24.33	4.12	28.2
CEBRACS^b
Alcohol effects
Before—eating less to get drunker	1.27	.68	35.3
Before—eating less to get drunk faster	1.24	.66	35.3
Before—skipping meals to get drunk faster	1.12	.50	35.4
During—eating less to get drunk faster	1.20	.59	35.2
During—not eating to get drunk faster	1.16	.56	35.3
During—eating less to get drunker	1.17	.58	35.3
During—not eating to get drunker	1.12	.51	35.3
Bulimia
Before—laxative use	1.04	.26	35.3
During—diuretic use	1.03	.24	35.3
During—laxative use	1.03	.26	35.3
After—diuretic use	1.03	.20	35.0
After—laxative use	1.03	.22	34.9
After—vomiting	1.04	.26	35.08
Dietary Restraint and Exercise
Before—exercising	1.50	.95	35.3
During—eating low calorie foods	1.21	.65	35.4
During—drinking low caloric alcoholic drinks	1.34	.86	35.51
After—eating low calorie foods	1.27	.72	34.9
After—exercising	1.60	1.03	34.9
After—eating less	1.18	.61	34.9
Restriction
Before—skipping meals	1.16	.57	35.3
After—skipping a day of eating	1.06	.33	34.9

a Descriptive statistics are based on all available values for each variable.

b CEBRACS item-level means range from 1-never to 5-almost all the time.

Table 2. Scale correlations and reliabilities missForest model with missing indicator.^a

Variables	Mean^b	1	2	3	4
1. Alcohol effects	NA	(.94)
2. Bulimia	NA	.80***	(.89)
3. Dietary restraint and exercise	NA	.75***	.77***	(.87)
4. Restriction	NA	.89***	.91***	.85***	(.63)
5. General factor	NA	.88***	.90***	.85***	.99***

a n = 1,149. Cronbach’s alphas for each sub-scale is located on the diagonal in parentheses.

b Means of CEBRACS latent variables are zero with a single time point.

*** p ≤ .001.

The association between race (i.e. White, Black, and Asian) and engagement in weight conscious drinking behaviors was assessed. A chi-square test of independence indicated that race was significantly related to student engagement in weight conscious drinking behaviors, X² (2, N = 694) = 11.9, p = .003. Post-hoc comparisons of race by student engagement in weight conscious drinking behaviors revealed that significant associations were observed among White (X² (1, N = 735) =5.6, p = .02) and Black (X² (1, N = 735) =10.5, p = .001) students, but not Asian (X² (1, N = 735) =.15, p = .69) students. The association between Hispanic ethnicity and engagement in weight conscious drinking behaviors was also examined. Results suggested Hispanic ethnicity was not associated with student engagement in weight conscious drinking behaviors, X² (1, N = 725) = .07, p = .79. A chi-square test of independence was conducted to examine the relation between gender and baseline weight intentions, irrespective of engagement in weight conscious drinking behaviors. The relation between these variables was significant, X² (3, N = 1,149) = 137.4, p < .001. Females were more likely than males to intend to lose weight or stay the same weight, whereas males were more likely than females to want to gain weight or to not try to do anything about their weight. The association between gender and weight intentions remained significant across students reporting weight conscious drinking behaviors (X² (3, N = 344) = 47.6, p < .001) and students who did not report weight conscious drinking behaviors (X² (3, N = 805) = 95.2 p < .001). A larger percentage of males engaging in weight conscious drinking behaviors wanted to stay the same weight, as compared to males who were not engaging in weight conscious drinking behaviors. Females engaging in weight conscious drinking were more likely to report wanting to lose weight and less likely to report not trying to do anything about their weight than female counterparts who did not engage in weight conscious drinking (see Table 3).

Table 3. Chi square test of independence between gender and weight intention by engagement in weight conscious drinking behaviors (n = 1,149).

	Weight conscious drinking behaviors not reported (n = 805)			Weight conscious drinking behaviors reported (n = 344)
	Male n (%)	Female n (%)	X^2	Male n (%)	Female n (%)	X^2
Weight intentions			95.2***			47.6***
Lose weight	111 (42.7)	331 (60.7)		54 (43.2)	156 (71.2)
Gain weight	76 (29.2)	28 (5.1)		33 (26.4)	8 (3.7)
Stay the same weight	34 (13.1)	112 (20.6)		24 (19.2)	42 (19.2)
I am not trying to do anything about my weight	39 (15)	74 (13.6)		14 (11.2)	13 (5.9)

*** p ≤ .001.

Measurement model fit was examined via a first-order confirmatory factor analysis (CFA) and results suggested adequate model fit (χ2 [186] = 327.385, p ≤ .001; CFI = .99, TLI = .99, RMSEA = .03, WRMR = 1.01). Since a significant χ2 test can reflect sensitivity to a larger sample size, model evaluation was determined by adequate cutoff criteria for relative fit indices.²⁹ For model identification and scaling purposes, the model was specified with the factor variance for Restriction fixed to zero (when the general weight conscious drinking factor was specified) and the first factor loadings for the second-order factors fixed to one. A summary of model fit indices is provided in Table 4.

Table 4. Summary of model fit indices for missForest model.

Model test	χ^2	df	CFI	TLI	WRMR	RMSEA
Measurement Model	327.385	186	.99	.99	1.01	.03
Hypothesized Model 1	779.87	283	.98	.98	1.58	.04

a χ²-values for the measurement and structural models are significant at p ≤ .001.

Weight conscious drinking factors and BMI

Structural modeling results indicated that none of the weight conscious drinking factors (i.e. Alcohol Effects, Bulimia, Dietary Restriction and Exercise, and Restriction) had a statistically significant positive or negative direct effect on BMI at 7-month follow-up; thus, indicating weight maintenance.

Gender differences and weight conscious drinking factors

Structural equation model results indicated males had a significant negative direct relationship with Alcohol Effects (β = −.15, SE = .05, p = .005). Gender was not found to have a significant effect on any of the remaining weight conscious drinking factors (i.e. Bulimia, Dietary Restriction and Exercise, and Restriction subscales). Structural model results with significant path coefficients are shown in Table 5.

Results for the missForest single imputation model including CEBRACS and BMI at follow-up missing indicators indicated there were no significant differences between students who skipped questions on the CEBRACS and those who did not nor between students who were present for anthropometric measurement at follow-up and those who were not (see Table 5).

Table 5. Standardized path coefficients.^a

Model test	Available-case analysis (n = 999)	missForest (n = 1,149)	missForest with missing indicator(n = 1,149)
Alcohol effects^b	−.04 (.12)	−.09 (.10)	−.09 (.09)
Bulimia^b	−.12 (.19)	.06 (.13)	.07 (.10)
Dietary restriction and exercise^b	−.03 (.10)	−.07 (.08)	−.07 (.06)
Restriction^b	.13 (.27)	−.11 (.21)	.09 (.19)
Gender^b	.01 (.01)	.02 (.01)	.01 (.02)
BMI at baseline^b	.93*** (.003)	.90*** (.004)	.91*** (.005)
Gender^c	−.16** (.05)	−.11* (.05)	−.15** (.05)
BMI at follow-up missing indicator^b			−.02 (.01)
CEBRACS missing indicator^b			−.001 (.01)

a Standard errors are in parentheses.

b Body mass index at follow-up is the dependent variable.

c Alcohol effects are the dependent variable.

*** p ≤ .001.

** p ≤ 0.01.

* p ≤ 0.05.

Discussion

This study focused on examining the effect of gender on weight conscious drinking dimensions and the effect of these dimensions on BMI at 7-month follow-up across a freshman U.S. college student cohort. Findings from this study highlight there are more similarities than differences in weight conscious drinking among freshman male and female students. These findings corroborate previous qualitative research suggesting that both genders engage in weight conscious drinking behaviors.^16,30 However, this study contributes to the theoretical literature on weight conscious drinking by establishing gender differences are present across the Alcohol Effects weight conscious drinking dimension. Student behaviors intended to enhance alcohol effects comprised the only weight conscious drinking dimension significantly related to gender with males being less likely to engage in Alcohol Effects behaviors as compared to female counterparts. As such, our study furthers weight conscious drinking theory by establishing that while male and female college students both engage in weight conscious drinking behaviors across all four weight conscious drinking dimensions (i.e., Alcohol Effects, Bulimia, Dietary Restriction and Exercise, and Restriction), females are significantly more likely than males to engage in Alcohol Effects behaviors. Our study findings also corroborate prior research indicating college females are more likely to engage in compensatory strategies to enhance the psychoactive effects of alcohol.^4,8,13 It is possible that gender differences across other weight conscious drinking dimensions reported in the literature stem from inconsistencies in the operationalization of weight conscious drinking. For instance, one study using latent variables to capture compensatory behavior dimensions did not specifically indicate compensatory behaviors were performed exclusively to compensate for alcohol-related calories.⁹ Other studies operationalized weight conscious drinking with a single item asking students to report how many times in the past 30 days they restricted energy intake before drinking alcohol.^4,31 Given this study holistically conceptualized weight conscious drinking to include the manifestation of disordered eating behaviors (among both college males and females) to enhance the psychoactive effects of alcohol, individually or in combination with behaviors aiming to offset the consumption of alcohol-related calories, our findings highlight the theoretical importance of the inclusion of all four weight conscious drinking latent dimensions in future research examining the weight conscious drinking phenomenon.

Findings from this study also suggest there was no significant change in BMI at 7-month follow-up among the college freshman cohort under examination (which reported engaging in alcohol-related compensatory behaviors). A baseline examination of student weight intentions by gender and engagement in weight conscious drinking behaviors revealed that a larger percentage of males engaging in weight conscious drinking behaviors wanted to stay the same weight, as compared to males who were not engaging in weight conscious drinking behaviors. Females engaging in weight conscious drinking were more likely to report wanting to lose weight and less likely to report not trying to do anything about their weight than female counterparts who did not engage in weight conscious drinking. These findings present implications for college alcohol abuse prevention professionals and treatment providers who may not perceive college males at greater risk for use of weight conscious drinking behaviors as a means of weight control. Clinicians, deans of students, members of student affairs, and judicial affairs personnel all have varied and pervasive interest in reducing alcohol use by college students. Research that reveals the multiple reasons underlying gender differences in weight conscious drinking behaviors among college students and thus better informs intervention programs is warranted. Programs that target alcohol use potentially by addressing parallel risk factors, in this case weight conscious drinking behaviors, allows for a multi-pronged intervention approach. These findings also bear valuable implications for the framing and delivery of preventive weight conscious drinking messages disseminated by college campus health professionals. Prevention programs should be tailored to address female tendencies to engage in compensatory strategies to enhance the psychoactive effects of alcohol and the deleterious effects of engaging in alcohol-related compensatory behaviors should be emphasized over its motivation as a potential weight management strategy, especially for males.

This study was not without limitations. Our inclusion criteria consisted of factors, identified by the study investigators, that would place college freshmen at risk for unhealthy diet and exercise behaviors, which limits the generalizability of our study results. Nonetheless, on average, our final study sample was composed of normal weight students with representative percentages of racial/ethnic student minorities.³² Given the stigmatized nature of disordered eating behaviors, there was a large percentage of missing data across CEBRACS items; thus, entailing greater potential for biased parameter estimates. In order to account for potential bias related to missing data, the missing data mechanism was assessed via the missing-indicator method and various model results were compared for triangulation of results. The missForest single imputation model including CEBRACS and BMI at follow-up missing indicators suggested there were no significant differences between students who skipped questions on the CEBRACS and those who did not skip questions nor between students who were present for BMI measurement at follow-up and those who were not present for BMI measurement.

Further qualitative research is warranted to examine the relationship between female gender and Alcohol Effects. It is not entirely clear why females are more inclined to engage in behaviors that enhance the psychoactive effects of alcohol, as compared to male counterparts, given their lower alcohol metabolism rate. In addition, future longitudinal research with three or more time-points is warranted to test for nonlinear weight conscious drinking trajectories and to obtain enhanced precision of parameter estimates. Moreover, the longitudinal stability of the gender effects observed in this study should also be further examined, in future research, with large representative college samples.

Authors’ contribution

Authors GC, TEB, WLL, and AEM designed the study, SEC, MDO, WZ, DS, and AEM collected data, GC conducted the statistical analysis, GC wrote the first draft of the manuscript. All authors contributed to and have approved the final manuscript.

Ethical standards disclosure

This study was conducted according to the guidelines laid down in the Declaration of Helsinki. The Institutional Review Board (IRB) at the University of Tennessee approved research activities at the University of Tennessee, West Virginia University, South Dakota State University, Syracuse University, and University of Maine. University of Florida, Auburn University, and Kansas State University Institutional Review Boards provided approval for research on their respective campuses. Written informed consent was obtained from all subjects.

Conflict of interest disclosure

The authors have no conflicts of interest to report. The authors confirm that the research presented in this article met the ethical guidelines, including adherence to the legal requirements, of the United States of America and received approval from the Institutional Review Boards of University of Tennessee, University of Florida, Auburn University, and Kansas State University.

Additional information

Funding

Funding for this study was provided by USDA Grant 2014-67001-21851. USDA had no role in the study design, collection, analysis or interpretation of the data, writing the manuscript, or the decision to submit the paper for publication.]