Big drinkers: How BMI, gender and rules of thumb influence the free pouring of wine

Highlights

• An observational study examines how BMI and gender might lead to the overpouring of wine.

• For men, the amount poured was positively associated with BMI. BMI did not affect women's pours.

• Using the “half glass” rule-of-thumb reduced the volume poured by over 20% for both men and women.

• The half-glass rule-of-thumb substantially reduced pours by men at high BMI levels.

• Using the “half glass” rule may effectively curb overpouring despite non-standard glass sizes.

Abstract

Background

This research examines free pouring behavior and provides an account of how Body Mass Index (BMI) and gender might lead to the overpouring, and consequently the overconsumption of wine.

Methods

An observational study with young adults investigated how BMI and gender affect free-pouring of wine over a variety of pouring scenarios, and how rules-of-thumb in pouring affect the quantities of alcohol poured by men and women across BMI categories.

Results

For men, the amount poured was positively related to BMI. However, BMI did not affect pours by women. The use of the “half glass” rule-of-thumb in pouring reduced the volume of wine poured by over 20% for both men and women. Importantly, this rule-of-thumb substantially attenuated the pours by men at high BMI levels.

Conclusion

Increasing awareness of pouring biases represents an early and effective step toward curbing alcohol consumption among men, and especially those who are overweight. Additionally, using a simple “half glass” rule-of-thumb may be an effective way to curb overpouring, despite non-standard glass sizes.

Introduction

A large body of research on alcohol consumption is based on self-reports of drinking behavior that assume participants’ accurate recall of past drinking behavior and ability to estimate standard drink sizes. Fewer studies examined real-time drinking behavior and behaviors preceding drinking such as individuals’ free-pouring of alcohol. This research addresses this knowledge gap by examining real-time alcohol pouring, as an antecedent of drinking behavior (Goddard, 2007, White et al., 2003, White et al., 2005). The scope of the present research is to examine how individual variables such as gender and BMI affect the amount of alcohol poured and ultimately consumed. In addition, we investigate how using arbitrary simple rules-of-thumb such as the “half glass rule” affects alcohol pouring behavior.

Assessing a standard drink size can pose challenges to consumers in situations or settings where free pouring is the norm. Goddard (2007) suggested that misperceptions of alcoholic strength and volume during pouring are major factors contributing to drinking problems in United Kingdom. The volume of beer and spirits consumed may be more accurately measured compared to that of wine. That is, beer is served from standardized single serving containers, cans, or full glasses (Devos-Comby & Lange, 2008), and spirits are generally measured in drinking establishments. In contrast, assessing wine consumption can be particularly challenging as wine is often self-poured at dinners, parties, and receptions. In addition, the amount poured may vary widely because the capacity of most wine glasses greatly exceeds one serving of alcohol, and the wide variety of glass shapes and sizes make poured volumes difficult to estimate relatively to a standard drink size metric (Goddard, 2007, Kerr et al., 2009, Walker et al., 2013). To date, most of the research on alcohol pouring has focused on the effect of external cues on the quantities of alcohol poured by bartenders (Kerr et al., 2008, Wansink and van Ittersum, 2005) and fewer studies have focused on the individuals’ self-serving of alcohol beverages (Gill and Donaghy, 2004, Banwell, 1999).

This research addresses this important gap in the literature and investigates how gender, BMI, and the use of rules-of-thumb relate to free-pouring of alcohol in a variety of consumption settings. We focused on wine to examine pouring biases because wine is consumed without ice or a mixer, and allows us to keep a constant % ABV (alcohol by volume) across participants.

Past research has found correlations between BMI and drinking quantity and/or frequency, however the results are mixed (Arif and Rohrer, 2005, Breslow and Smothers, 2005, French et al., 2010, Nies et al., 2012, Pajari et al., 2010). Some studies suggest that the quantity of alcohol consumed is positively related to BMI; consuming four or more drinks per day was associated with significantly higher BMI than consuming one drink a day (Breslow & Smothers, 2005). Also, binge drinkers and heavy drinkers had higher BMI than those who frequently consumed small amounts of alcohol (Arif and Rohrer, 2005, Breslow and Smothers, 2005). On the other hand, frequency of drinking alone was found to be negatively related to BMI. Frequent drinkers had lower BMI than participants who drank less frequently (Breslow & Smothers, 2005).

Other studies found the relationship between BMI and drinking to be moderated by gender. These studies also present mixed results. Some studies suggest a negative correlation for women between BMI and both the frequency and quantity of alcohol consumed (Colditz et al., 1991, Lahti-Koski et al., 2002, Liu et al., 1994, Rohrer et al., 2005); others suggest a positive correlation between BMI and the number of drinks consumed weekly (Smarandescu, Walker, & Wansink, 2014). For men, the findings are also inconclusive. Some studies found the quantity of alcohol consumed to be positively correlated with BMI (Arif and Rohrer, 2005, Breslow and Smothers, 2005, French et al., 2010, Lahti-Koski et al., 2002, Nies et al., 2012, Pajari et al., 2010), while others reported a negative correlation, or no correlation with BMI (Colditz et al., 1991, Liu et al., 1994, Smarandescu et al., unpublished manuscript, Williamson et al., 1987). Gearhardt and Corbin (2009) proposed that drinking quantity may differ with BMI levels because heavier individuals are likely to drink more in order to attain comparable blood alcohol concentration (BAC) levels. This may partly explain why men typically drink more than women (Berkowitz and Perkins, 1987, Johnson et al., 2004, Johnson, 1997, Lemle and Mishkind, 1989, McCreary et al., 1999, Perkins, 1992), as men have higher average weights than women.

Another factor that is likely to affect free-pouring and drinking behavior is individual perception of social norms. One of the reasons why men drink more than women is because drinking is perceived as more socially acceptable behavior for men than for women (de Visser & McDonnell, 2012). Research attributed the widespread use of alcohol among students to misperceptions of peer drinking norms (Baer and Carney, 1993, Baer et al., 1991, Perkins and Wechsler, 1996) and suggested that students overestimate the alcohol consumption of their peers (Perkins & Berkovitz, 1986), which leads to increases in their own consumption. Studies have also found that gender moderates the relationship between perceived social norms and drinking (Adams and Nagoshi, 1999, Lo, 1995). Studies showed that individual perceptions of same-gender drinking (Lewis & Neighbors, 2004) were more strongly correlated with drinking than perceptions of drinking behavior of a typical student (without reference to gender) in particular for women. They suggested that while consuming alcohol, women are more likely to adhere to social norms derived from same-gender comparison than men.

To date, no research has examined how BMI moderates the relationship between gender and alcohol pouring behavior, which is an important antecedent of alcohol consumption. We expect to find gender differences in pouring behavior based on the fact that men drink more often and more heavily than women. Hence, on average, men would pour larger volumes than women. It is also expected that men are more likely to drink to attain a certain BAC level; therefore, men would pour larger volumes at higher levels of BMI. In comparison, women are expected to pour less than men and more constant across BMI levels for two reasons. First, drinking is still viewed as less socially acceptable for women than it is for men (de Visser & McDonnell, 2012), and given that self-pouring alcohol has high visibility in social situations, women would be more likely to monitor their pour volumes. Second, consistent with research that found that gender moderates the relationship between perceived social norms and drinking (Adams and Nagoshi, 1999, Lo, 1995), women are more likely to engage in same-gender comparisons than men (Lewis & Neighbors, 2004) and comparisons with an average pour would result in less variance in pouring across scenarios. In conclusion, pour amounts for women are expected to increase with BMI at a smaller rate than for men.

Moreover, individuals’ pouring behavior may also be influenced by the use of certain rules-of-thumb in pouring. For instance, wine connoisseurs suggest that glasses should be filled about one-third of the way for red wines, halfway for whites, and three-fourths of the way for sparkling wines. Individuals who regularly adhere to these rules-of-thumb are expected to adjust their pour amounts accordingly and pour less than individuals who do not use rules-of-thumb. We also pose that the use of rules-of thumb in pouring will interact with gender and BMI. For women, who are likely to pour more consistent amounts across BMI levels, this would imply a general downward shift in pour amounts. For men, in addition to a downward shift in the amount poured, the use of a rule-of-thumb would likely attenuate the effect of BMI on pouring, as pour amounts are expected to increase along with BMI for men.

A quasi-experimental study examined wine-pouring behavior for a young adult college population (students and staff members) and correlated quantity poured as a function of gender, BMI, and use of rules-of-thumb. Wine was chosen as our target beverage in this study because it is free-poured in different consumption contexts and in glasses of different shapes and sizes. Notably, wine is consumed in pure form, and we could control its % APV, allowing us to solely focus on differences in pouring behavior across gender and BMI. Various pouring scenarios were used to control for familiarity with any particular set-up.

Although it is important to realize that pouring behavior does not necessarily equate to drinking behavior, based on a significant body of research in nutrition and obesity, which shows that serving behavior is correlated with food intake and overeating (Wansink, 2004, Wansink et al., 2005, Wansink and van Ittersum, 2007, Wansink et al., 2006), we anticipate wine-pouring behavior to provide an effective proxy for wine consumption.

After getting approval from the university Institutional Review Board, a convenience sample of 74 college students and staff of legal drinking age was recruited. Participants were recruited based on the criteria that they drank at least one glass of wine per week. After filling out consent forms, participants were directed to different pouring stations and instructed to pour “as much wine as they would normally pour into a glass in one setting”. They did not drink the wine. This pouring task was repeated over sixteen scenarios for each participant. Participants were instructed to consider each pour as a one-time individual serving, which was unconnected to previous pours. The tables used for pouring contained scales that measured the weight of the empty and filled glass when the glass was placed on the table. To control for the possibility of a demand artifact, in addition to the wine pouring tables, individuals were directed to other tables where they served themselves with pasta and apple sauce in different plate sizes. Participants did not eat the food. Sample statistics are shown in Table 1.

In order to derive findings that would be meaningful and generalizable considering the wide range of wine glass shapes and sizes existing in the marketplace, in this study we provided participants with a variety of pouring scenarios. Each pouring station presented a different context, and the set-ups were comprehensive of most common pouring situations. Across scenarios, individuals poured different types of wine (red or white), from bottles with different fullness levels, into different glass sizes (small, medium, large) and shapes (narrow, standard, with and without stem, ornate, or simple), and in the presence or absence of water pitchers. To further control for familiarity with a particular pouring position, participants were also instructed to pour from different positions (standing, sitting at a table, holding the glass, or having the glass placed on the table), and these positions were also randomized across scenarios. All sixteen pouring scenario were presented in random order to each participant; each scenario featured a unique combination of variable levels. The set of sixteen scenarios, out of a possible 10,368 combinations, allowed us to empirically control for the effects of each of the attributes and attribute levels in our model.

This comprehensive design allowed us to control for the effects of irrelevant contextual variables and individual familiarity with a setting, and focus on the effects of gender, BMI, and use of rules-of-thumb in pouring. Following the pouring task, the weights of the pours were recorded and participants stepped on a scale that measured their body weight. Next they completed a survey that asked them to report their height, demographics, and whether they used rules-of-thumb in wine pouring (see Table 1). Sixty-six percent of the individuals in our sample indicated that they were likely to use a rule-of-thumb when pouring wine (i.e. a score of greater than 5 on a 9-point scale anchored by “not at all likely” and “very likely”). The rules-of-thumb measures in pouring included, “I usually pour wine into a glass until there is a two finger gap between the wine and the top of the glass,” “When pouring wine to myself, I usually fill my glass to the top,” and “A person should never pour more than a half of glass of wine at one time.” No individuals reported that they were likely to use more than one rule-of-thumb. The “half glass” rule was used by 71% of the subjects reporting use of a rule-of-thumb and was therefore used in our analysis.

We modeled the amount poured, in grams, using a random parameters model, controlling for all the components of the sixteen different pouring scenarios and allowing for both observed and unobserved heterogeneity across subjects. The observed heterogeneity was specified as a function of the individual factors, gender, BMI, and the “half glass” rule-of-thumb, as well as all two and three-way interactions.

The parameter estimates for the hierarchical variables – the individual characteristics, appear in Table 2. Given the modeled interactions among the three individual factors, the effects γ₁–γ₆ are all conditional ones, and must be carefully interpreted. These effects are presented in Fig. 1 across the ranges of BMI for the males and females in our sample. A breakdown of the sample by gender and BMI appears in Table 3. Since we are interested in investigating differences in pour amounts at particular focal values of BMI, a spotlight analysis is particularly suited for interpreting our findings (Spiller, Fitzsimons, Lynch, & McClelland, 2013). Specifically, the simple effects of gender (i.e. male), the rule-of-thumb (RofT), and the interaction between the two (Male x RofT), which are all conditional on BMI, can be tested for significance at each focal value of BMI.

The first column of estimates in Table 2 corresponds to BMI being mean-centered at the mean levels of BMI for both the males and females in our sample. Columns 2, 3, and 4 correspond to centering BMI at the focal levels of the midpoint of the normal range (21.75), the normal/overweight cutoff value (25.00), and the overweight/obese cutoff value (30.00), respectively.

The most commonly used rule-of-thumb was the “half glass” rule (47%), followed by “to the top” (15%), and the “two finger gap” (4%) rules. Thirty-four percent of the sample did not use a rule-of-thumb. There was no significant difference in the use by rules-of-thumb by males vs. females. Because of the widespread use of the “half glass” rule and the less common usage of the other rules, subsequent analyses focus only on the former.

When the “half glass” rule was not used, males poured more than females for all three focal values of BMI. This effect can be seen graphically in Fig. 1, where the solid black line has been shown to be significantly above the solid gray line at all three focal values of BMI. Without the “half glass” rule, at the midpoint of the normal range of BMI, males poured 9% more than females. At the normal/overweight cutoff, that percentage increased to 19%. Males poured 34% more than females at the overweight/obese cutoff.

The estimates for the “half glass” rule (γ₃) indicate that females poured less when they were using the “half glass” rule than when they were not. This effect can be seen in Fig. 1 – the dotted gray line is lower than the solid gray line. At the midpoint of the normal BMI range, females poured 20% less when they used the “half glass” rule than when did not use the rule. This percentage dropped slightly to 19% at the overweight/obese cutoff.

The interaction effect between gender and the use of the “half glass” rule (γ₅) is shown to be non-significant when BMI is at the midpoint of the normal range (see Table 2, column 2). However, when BMI is at the cutoff between normal and overweight, the interaction is significantly negative at the 0.05 level (Table 2, column 3). When BMI reaches the overweight/obese cutoff, the interaction is negative and significant at the 0.001 level. This effect highlights the magnitude of the impact of the “half glass” rule on pouring behavior of males and females, visible in Fig. 1. The rule-of-thumb effect seems to be attenuated for females (shown as a slight reduction in the distance between the gray lines in Fig. 1) as BMI increases. For males, the rule-of-thumb effect increases (shown as an increase in distance between the black lines in Fig. 1) as BMI increases. Without using the “half glass” rule, males at the overweight/obese cutoff poured 31% more compared to 26% at the midpoint of the normal range.

For females who were not using the “half glass” rule, the effect of BMI on pour amounts (γ₂) is significantly negative (shown as the down-sloping solid gray line in Fig. 1), although the effect size is small, consistent with the expectation that females’ pour amounts do not increase with BMI. The effect of BMI on male pours is of a significantly higher positive magnitude than for females (γ₄) – the slope of the solid black line in Fig. 1 is steeper than the slope of the solid gray line. Males who were not using the rule-of-thumb poured 19% more at the overweight/obese cutoff than at the midpoint of the normal range.

The positive interaction effect between BMI and use of the “half glass” rule for females (γ₆), is graphically represented in Fig. 1, where the distance between the solid gray line and the dotted gray line in Fig. 1 is reduced as BMI increases. Finally, the positive impact of BMI for males is attenuated when the “half glass” rule was used (γ₇), effect indicated by the greater slope of the solid black line in Fig. 1 as compared to the dotted black line.