An experimental assessment of the effects of two alcoholic beverage health warnings across...

In recent years, alcohol-related problems on college campuses have been well documented. This research examines how two different alcoholic beverage health warnings placed on the label of a fictitious brand of beer influence alcohol-related risk perceptions, attitudes and intentions, and characterizations

of problem-drinking behaviors of binge and non-binge drinkers in Australia and the United States. The consumer welfare implications of our findings, which show that the dependent measures are influenced by both main and interaction effects, are discussed.

**********

In the past decade, university administrators and government policymakers have become increasingly concerned about excessive alcohol consumption and binge drinking by college students. Studies have reported that in the United States, alcohol has been involved in 98% of college rapes, 95% of violent crime on campus, and two-thirds of college student suicides (CASA Commission 1994). Positive associations between excessive consumption of alcohol by students and increased incidences of traffic fatalities, unplanned sexual activity, physical and sexual assault, unintentional injuries, physical and cognitive impairment, and poor academic performance also are reported (Hanson and Engs 1992; Presley et al. 1993; Wechsler and Issac 1992). In fact, a recent survey found that college presidents perceive binge drinking as the most serious problem on their campuses today (Center for Science in the Public Interest 2000).

Recognizing the huge social, health, and monetary costs associated with alcohol abuse, the U.S. government has required a warning label on all alcoholic beverage containers distributed and sold in the United States since 1989. This government-mandated statement warns that women should not drink during pregnancy and that alcohol consumption impairs your ability to drive a car or operate machinery. Prior research (Andrews, Netemeyer, and Durvasula 1993) has suggested that these two very specific consequences associated with alcohol consumption are relatively well known. Consequently, the use of new or rotating warnings may have a greater opportunity for garnering attention and increasing effectiveness (Andrews and Netemeyer 1996; Hilton 1993).

Concerns associated with excessive consumption of alcohol, especially by young adults, are not limited to the United States. Government policymakers, healthcare professionals, and university administrators are among those in other nations who are concerned about the monetary, social, and health consequences of excessive alcohol consumption. Consequently, governments from several countries around the world are now considering, or have recently considered, mandating the placement of a health warning statement on all alcohol beverage containers. A health warning statement placed on a package label is considered by policymakers to be an effective means by which to enhance consumers' awareness of potential problems associated with alcohol consumption. An enhanced awareness of the possible negative consequences of alcohol consumption may potentially modify attitudes and behaviors.

RESEARCH OBJECTIVES

The purpose of this study is to examine the effects of two health-warning statements that address the potential negative effects of alcohol consumption in college-aged consumers. One statement is the warning currently mandated by the U.S. government. The other statement is a variation of a warning recently considered by the Australian New Zealand Food Authority (ANZFA) that explicitly identifies alcohol as a drug. The influence of these health warnings on consumers classified either as binge or non-binge drinkers is considered across two countries (the U.S. and Australia). Specifically, we address three sets of research questions: 1) Does the type of health warning presented on an alcohol beverage container influence consumers' attitudes toward getting drunk, the anticipated social benefits of beer consumption, and the perceived negative health consequences of beer consumption? What influence will binge-drinking status and country of residence have on these alcohol-related attitudes? 2) What effects do the tw o different health-warning statements have on the perceived risk of various alcohol-related behaviors (e.g., driving after drinking; drinking when pregnant)? Are these effects also influenced by binge drinking status and country of residence? 3) Does the type of health warning statement presented on an alcohol beverage container influence consumers' beliefs regarding what constitutes "problem-drinking" behavior? Does the influence of the health warning statement differ between binge drinkers and non-binge drinkers from different countries?

Although the primary focus of this research is on the effects of the two health-warning statements on the different types of dependent measures, it is also important to consider the influence of both the consumer's country of residence and binge-drinking status. Sweeping changes in both communication and logistical capabilities in recent years has increased the number of consumers who cross national borders to obtain information and products. Thus, it is important to expand the study of consumer welfare issues to different countries (Bamossy, Belk, and Costa 1997). Binge drinking is not a problem unique to the United States; it is an important health and social issue in many other nations. While binge drinkers from different nations may share many common characteristics, no research has examined the influence of warnings on binge and non-binge drinkers across different countries. Because warning information may be processed and interpreted differently by these different consumer groups, it is important to con sumer welfare to begin to address this gap in the existing literature.

THEORETICAL BACKGROUND

Binge Drinking in the U.S. and Australia

On most college campuses in both the United States and Australia, binge drinking is a common occurrence linked to numerous negative consequences, including death. As noted by Treise, Wolburg, and Otnes (1999), public service announcement campaigns that encourage responsible drinking behavior have achieved only moderate successes; consequently there is a great need for additional insight into the problem of binge drinking. The severity of the consequences resulting from the excessive consumption of alcohol by teenagers and young adults is best illustrated by the following statistic: alcohol-related traffic accidents are the leading cause of death for U.S. consumers between the ages of 15 and 24 (National Committee for Injury Prevention and Control 1989; U.S. Dept. of Education 1994). Findings from Harvard's College Alcohol Study, which surveys students from 119 different colleges in the U.S. (Wechsler et al. 1999), show that 44% of U.S. college students (51% of males and 40% of females) are categorized as bing e-drinkers. (1) In addition, more than 20% of the students were categorized as frequent binge drinkers, and these students were more than twenty times more likely than non-binge drinkers to have driven a car after drinking, engaged in unplanned sexual activity or not used protection when having sex, become involved with campus police, been hurt or injured after drinking, damaged property, or missed classes and fallen behind in schoolwork (CSPI 2000). Given such findings, it is not surprising that college presidents now perceive binge drinking as the most serious problem facing their campuses.

While statistics related to binge drinking on college campuses are not as readily available for Australian college students, there is little reason to believe that the problem is any less severe in Australia. Studies have shown that Australians have the highest rate of alcohol consumption in the English-speaking world, and that Australians are the perhaps the world's most prodigious group of binge drinkers in terms of per capita alcohol consumption (Reuters 1994). Others have noted that alcohol is an indispensable element of the male bonding experience ("mateship") (National Rural Health Alliance 1998). One recent study (Kerr, Fillmore, and Marvy 2000) shows that per capita consumption of both beer and ethanol has been greater for Australians than for Americans every year since 1953 (which was the first year statistics were available). It has been reported that by the age of 16, one-third of Australian consumers are already binge drinkers. Thus, binge drinking is a serious social and medical problem confronti ng consumer welfare advocates, public policymakers, and healthcare workers in both the United States and in Australia. However, whereas the U.S. government mandates the placement of warning statements on alcohol beverage containers, the Australian government does not.

Alcohol Warning Labels

The following nine countries currently have laws mandating health-warning labels on alcohol beverage containers: Brazil, Columbia, Costa Rica, Ecuador, Honduras, Mexico, South Korea, the United States of America, and Zimbabwe. Several other countries--including Australia, Canada, France, Japan, New Zealand, Taiwan, Thailand, and South Africa--are currently addressing or have recently addressed this issue. However, health-warning labels are not currently used in any European country. A potential health-warning label recently under consideration by the Australian New Zealand Food Authority (ANZFA) is of particular interest to our research. In April of 1998, the Society Without Alcohol

Trauma requested an amendment to the Food Standards Code to mandate the inclusion of a health-warning statement on all alcohol beverage containers. Whereas the current U.S. health warning identifies specific potential consequences of alcohol consumption, the proposed Australian warning was more general in nature; it simply indicates that the product contains alcohol and that alcohol is a dangerous drug.

While some nations are presently considering mandated health warnings on alcohol beverage containers, some public policymakers and healthcare professionals within the U.S. are now questioning the effectiveness of the U.S. mandated warnings. There are several reasons why the effectiveness of the health warning currently used in the U.S. is being reexamined, particularly in terms of its effects for binge and frequent drinkers. First, since the same health warning has been in use since 1989, frequent drinkers may be habituated to it and thus ignore the message (Andrews and Netemeyer 1996). Second, since consumers have been exposed to these specific health warnings for many years, it seems likely that awareness of the dangers associated with drinking when pregnant and driving a car or operating machinery while under the influence of alcohol has peaked; all accessible target consumers have been reached (Greenfleld 1999; Hankin, Sloan, and Sokol 1998).

While these messages related to drinking and driving and drinking when pregnant are clearly important ones, many users are familiar with these themes. Consequently, these statements may no longer have the same positive persuasive impact that they had when first introduced. Other warning label content may be both more persuasive and compelling. Other messages that provide novel information, and thus present alcohol beverages in a different cognitive frame, may be more effective in conveying consumption risks and influencing perceptions about problematic levels of consumption.

For example, consider the alcoholic beverage health warning that was under consideration by ANZFA stating, "Alcohol is a dangerous drug." This message may potentially modify consumers' cognitive frames and perspectives regarding alcoholic beverages. Many consumers are unlikely to consider beer, wine coolers, and other alcoholic products to be "drugs." Conveying this information in a warning may add informational links to the conceptual node in memory for "beer" (Anderson 1983; Collins, and Loftus 1975). That is, associative memory network concepts can be used to postulate consumer response to information linking alcoholic beverages and drugs. Thus, identifying beer as a product that contains alcohol, a drug, may evoke concepts associated with negative long-term consequences, such as "addiction," and associations with substances for which consumption risks are more commonly accepted such as cocaine or amphetamines. Simultaneously, it may also detract from beer's advertised image as a fun, social, and innocuous product. Use of a health warning that explicitly identifies alcohol as a drug may potentially have the greatest effect on the risk perceptions and attitudes of the heaviest users of alcoholic beverages. These users may be most likely to ignore and/or discount the current U.S. warnings that use common and wellknown themes like drinking and driving (Andrews et al. 1991).

HYPOTHESES

Based on the above discussion, a series of hypotheses are proposed to address the research issues; three sets of dependent variables important to public policymakers, healthcare workers, and consumer welfare advocates are considered. First, we examine how different health-warning statements influence alcohol-related attitudes and purchase intentions across binge-drinking status and country. These attitudinal constructs include perceived social and health benefits and attitude toward getting drunk (Weschler et al. 1999). Such drinking-related attitudes and intentions are important to consider because prior research has identified both attitudes and intentions as antecedents that influence one's behaviors (Ajzen and Fishbein 1980).

The type of health warning statement presented on an alcoholic beverage label is likely to influence attitudes and intentions. A warning statement that explicitly identifies alcohol as a drug, compared to one that discusses the negative, but well-known consequences of drinking and driving and drinking when pregnant, is expected to activate more serious and negative drug-related associations, and thus should have a more negative influence on alcohol-related attitudes and intentions.

Alcohol-related attitudes and intentions are also likely to be influenced by the consumers' country of residence. Given the high rate of alcohol consumption and its widespread acceptance within Australia (Kerr et al. 2000; Reuters 1994), Australians are expected to have more positive drinking-related attitudes and intentions than Americans. Similarly, theory and prior research (e.g., Ajzen 1985; Wechsler et al. 1999) has shown that, generally, there is a positive correlation between attitude and behavior. Thus, binge drinkers should have more favorable attitudes and intentions regarding the consumption of alcohol than non-binge drinkers. Thus, the following hypothesis is proposed:

H1: The type of health warning to which consumers are exposed, their country of residence, and their binge-drinking status will influence consumers' alcohol-related attitudes and intentions. Specifically, a) presenting the health warning statement that identifies alcohol as a drug will result in more negative alcohol-related attitudes and intentions. In addition, b) Americans and c) non-binge drinkers will have more negative attitudes.

Based on the rationale offered above, the perceived risk associated with a variety of alcohol-related behaviors and consumers' beliefs regarding "problem-drinking behaviors" are also expected to be influenced by the main effects of country, binge drinking status, and health-warning message. Researchers and policymakers have been concerned about the level of perceived risk associated with a variety of behaviors (e.g., driving after drinking, drinking when pregnant, risk of DWI/DUI). (2) Prior research has shown that certain consumers may underestimate their personal risk in response to a warning because they view themselves as more skilled than the average consumer (Slovic, Fischoff, and Lichenstein 1980; Svenson, Fischoff, and Macgregor 1985). For example, some consumers may perceive less risk than others from driving after drinking or drinking when pregnant because they believe they are able to "handle" their alcohol better than other consumers. This phenomenon, termed personal immunity, may be extended to b inge drinkers' perceptions of risk when viewing a warning.

H2: The level of perceived risk associated with alcohol-related behaviors (e.g., driving after drinking, drinking when pregnant, risk of DWI/DUI) are expected to be influenced by the type of health warning statement to which consumers are exposed, binge-drinking status, and country of residence. Specifically, a) consumers presented the mandated U.S. warning focusing on not drinking when pregnant or before driving will perceive less risk than those presented with the warning identifying alcohol as a drug. In addition, b) Australians and c) binge drinkers will perceive less risk.

In this study, perceived risk is addressed across increasing levels of consumption (1-2 beers, 3-4 beers, 5-7 beers, etc.). We examine the intuitive prediction that greater consumption amounts lead to increased levels of perceived risk. In addition to the expected main effects of consumption amount, country, drinking status, and warning, we postulate that drinking status and country will moderate effects of consumption amount on risk perceptions. Consistent with the personal immunity theory of risk, binge drinkers should perceive significantly less risk at lower levels of consumption (i.e., 1-2, 3-4 beers). However, at higher levels of consumption, the difference between the risk perceptions of binge and non-binge drinkers should decrease. That is, binge drinkers are expected to perceive less risk in general, but at some (high) level of consumption even binge drinkers will realize that risks associated with behaviors such as driving after drinking or drinking when pregnant are high. This rationale suggests an ordinal interaction between drinking status and consumption amount.

We predict that the respondents' country of residence will moderate the effect of consumption amount on risk perceptions related to driving and pregnancy. Warning statements presented on alcoholic beverage containers in the U.S. have emphasized the dangers associated with both drinking when pregnant and drinking and driving for the past 12 years. National promotional campaigns have also emphasized the importance of not drinking and driving and have warned women to avoid alcohol consumption when pregnant. As a result of these campaigns, consumers in the U.S. are very aware of these risks. Consequently, we predict that at lower levels of consumption, Australians will perceive the risks associated with drinking and driving and drinking when pregnant to be lower than U.S. consumers, but this difference will no longer be significant as consumption levels increase. The main effect of consumption level on perceived risk is addressed in H3a while H3b and H3c present the predicted ordinal interactions.

H3a: The level of perceived risk associated with alcohol-related behaviors will increase as consumption amounts increase.

H3b: The effects of consumption amounts on perceived risk are moderated by binge-drinking status. Binge drinkers will perceive less risk at lower levels of consumption than non-binge drinkers, but risk differences between binge and non-binge drinkers will decrease at higher levels of consumption.

H3c: The effects of consumption amounts on perceived risk of drinking when pregnant and driving-related risks are moderated by country of residence. Australians will perceive less risk than Americans at lower levels of consumption, but risk differences between Australian and U.S. consumers will decrease at higher levels of consumption.

It is also important for policymakers, healthcare workers, and consumer welfare advocates to understand the factors influencing consumers' characterizations of "problem-drinking." The first step in recovery from the negative consequences of excessive alcohol consumption is recognition of a "drinking-problem" (Johnson 1980; Pittel 1985). Thus, before encouraging a consumer to reduce, or limit, his or her drinking to a healthy level, it may be necessary to understand what is perceived to be "acceptable" consumption behavior and what is perceived to be excessive consumption, or "problem-drinking." Binge drinkers may not recognize, or may not admit to themselves, the potential negative consequences associated with their drinking behavior. Binge drinkers are therefore less likely to perceive their drinking as problematic.

However, explicitly identifying alcohol as a drug may cause consumers to consider alcohol in a different, and more negative, light. This reevaluation may activate drug-related associations such as addiction and withdrawal, and thus cause binge drinkers to reconsider their behaviors. Information presented in the standard U.S. warning is unlikely to have the same powerful impact, especially if the consumer is not a woman considering pregnancy or is not someone who plans to consume alcohol and then drive or operate machinery. The influence of the different health warning statements is likely to interact with binge drinking status to influence consumers' identification of "problem drinking" behaviors. That is, binge-drinking status is expected to moderate consumers' responses to the different health-warning statements. Specifically, we expect the following:

H4: Binge drinking status will influence consumers' identification of problem drinking behaviors. That is, a) binge drinkers will perceive lower levels of problem drinking than non-binge drinkers across different consumption levels. However, this main effect will be moderated by the interaction between binge drinking status and type of health warning. That is, b) binge drinkers will perceive lower levels of problem drinking than non-binge drinkers when the U.S. health warning is presented. However, when the health warning states, "Alcohol is a Drug," differences between the perceptions of binge and non-binge drinkers will decrease.

PILOT STUDY

Based on per capita consumption estimates and existing literature (Kerr et al. 2000; Reuters 1994), Australian students were expected to be more favorably inclined toward the consumption of beer (and alcoholic beverages in general) than American students. To assess this premise, a small pilot test (utilizing students who did not participate in the main study) was performed in which several consumption-related variables were measured. In addition, procedures and dependent variables intended for use in the main study were assessed. In this pilot test, 45 undergraduate students from the U. S. and Australia were exposed to the mock-up of the beer bottle stimulus, but without any type of warning condition included. Subjects responded to items proposed as dependent variable measures for the main study. These measures were followed by consumption-related variables in the final section of the survey.

Alcohol consumption-related questions used categorical responses to assess how many times beer was consumed in the past 30 days, the number of beers consumed on occasions when beer was drunk, self-classification of drinking status (e.g., an abstainer, moderate drinker, heavy drinker), occasions in which five and ten beers had been consumed in a single sitting over the most recent two-week period, and the number of occasions and average number of drinks per occasion that had been consumed over the past 30 days. Measures for each of these consumption variables were drawn from the Harvard College Alcohol Survey (Wechsler et al. l999). (3) Results showed that Australian students scored somewhat higher on each of these consumption measures and the [chi square] value pooled across the measures was significant ([chi square] = 23.9, df 12; p < .05). In addition, measures of the multi-item scales all had acceptable reliabilities (i.e., [alpha] > .80) across both countries, and there were no problems encountered with t he procedures for either the Australian or U.S. students.

EXPERIMENT

Method

Study respondents were 274 students recruited from two universities, one located in the U.S. and one in Australia. All respondents were undergraduate students who voluntarily participated in this study. There was no significant difference in ages across the two samples; the average age of both the U.S. and the Australian respondents was 21 years. In this study, 48% of the respondents were male and 52% were female, and the binge drinker by country by gender cross-tabulation revealed no significant difference. Similarly, the overall percentage of respondents classified as binge drinkers did not differ across respondents from the two countries (p > .10).

Across the total sample of 274, there were 114 respondents classified as binge drinkers and 160 as non-binge drinkers, 168 U.S. and 106 Australian respondents, and 128 who were exposed to the standard U.S. warning and 146 to the experimental warning specifying that "Alcohol is a Drug."

Research Design and Procedure

Subjects were randomly assigned to experimental conditions and data were collected in a classroom setting. Upon entering the classroom, booklets containing the experimental stimuli and sets of dependent measures were randomly distributed to students. On the first page of the booklet, subjects were exposed to an actual-sized reproduction of a beer bottle with a fictitious brand name (Blackman Premium beer) and other information about the product (3.5% Alcohol by Volume; 12 fluid oz = 355m1) that was invariant across experimental conditions.

Two warning conditions were varied across the beer stimuli. The first condition was the standard warning required on alcoholic beverage labels in the U.S. stating: "GOVERNMENT WARNING: (1) WOMEN SHOULD NOT DRINK ALCOHOLIC BEVERAGES DURING PREGNANCY BECAUSE OF THE RISK OF BIRTH DEFECTS. (2) CONSUMPTION OF ALCOHOLIC BEVERAGES IMPAIRS YOUR ABILITY TO DRIVE A CAR OR OPERATE MACHINERY AND MAY CAUSE HEALTH PROBLEMS." The second experimental warning stated the following: "GOVERNMENT WARNING: THIS PRODUCT CONTAINS ALCOHOL. ALCOHOL IS A DRUG." As noted previously, this latter warning was based on a warning for alcoholic beverages that was being considered, but had not been used, in Australia.

The two warnings were tested for both binge and non-binge drinkers of beer and for subjects from the United States and from Australia. Binge drinking for beer was a measured variable and was based on the classification procedure used in Harvard's College Alcohol Studies in which drinking attitudes and behaviors of some 32,000 students from 140 colleges in the two most recent studies were examined (e.g., Wechsler et al. 1994, 1999, 2000). Following this classification procedure, subjects are asked two separate questions: "How many times have you had five or more 12 ounce beers in a row?" and "How many times have you had four or more 12 ounce beers in a row?" The time frame for each question is the past two weeks. Binge drinking status is then determined by combining responses with a question for gender. Binge drinking is defined as five or more drinks for a single occasion for a male and four or more drinks for a female (Wechsler et al. 1994, 1999). Respondents who report consuming less than these respective a mounts during the prior two-week period are classified as non-binge drinkers. This measure of binge drinking is the one most cited in current literature on the topic (Wechsler and Austin 1998).

Measures

Several types of responses to the experimental stimuli were assessed, including drinking-related attitudes and intentions, perceptions of risk across various levels of alcoholic beverage consumption, and perceptions of the level of problem drinking for different hypothetical consumption behaviors. All dependent measures used 7-point scales, and all multi-item scales were divided by the number of items in the scale so that results are consistent with a 1 to 7 scoring. Each of the sets of measures is described below.

Drinking-Related Attitudes/Intentions. After exposure to the experimental stimuli, all attitude and intention measures, including attitude toward getting drunk, perceptions of the social benefits associated with beer drinking, perceptions of health benefits, and purchase intentions, were presented. The attitude toward getting drunk was measured with a three item semantic differential scale that asked, "What is your attitude toward drinking enough to get drunk (By drunk we mean unsteady, dizzy, sick to your stomach)?" The following endpoints were presented: 1) Not Enjoyable--Enjoyable, 2) Unfavorable--Favorable, and 3) Negative--Positive. This measure was adapted from the Harvard Alcohol Study instrument (Wechsler et al. 1999), and coefficient alpha was .97. Five items were used to measure the perceptions of the social benefits from drinking beer, and coefficient alpha was .85. The items all used endpoints of "strongly disagree" and "strongly agree" and were as follows: "Drinking beer makes it easier to connec t with peers"; "When you drink beer you have more fun"; "Drinking beer facilitates sexual opportunities"; "Drinking beer makes it easier to deal with stress"; "Drinking beer gives people something to do." Three items were used to measure health benefits ("Do you consider the health benefits associated with drinking beer to be ... " with endpoints of Poor-Good, Negative-Positive, and Unfavorable-Favorable). Coefficient alpha was .96. These two items measured purchase intentions for the (fictitious) beer product mock-up used in the study: "Assuming you were going to buy a bottle of beer, would you be more likely or less likely to purchase the product, given the information shown?" (endpoints of "more likely" and "less likely"; reverse coded) and "How likely would you be to purchase the product, given the information shown on the bottle of beer?" (endpoints of "very likely" and "very unlikely"; reverse coded). The correlation between these two items was .63.

Perceived Risk Measures. Next, measures of perceived risk were presented; these measures assessed respondents' perceptions of the level of risk associated with performing various behaviors and/or conditions after consuming different amounts of beer. Specifically, the risks associated with drinking when pregnant, drinking and driving, doing something you later regretted after drinking, performing poorly on a test or important project because of drinking, and being arrested for driving while intoxicated or driving under the influence (DWI/DUI). These categories were drawn from those used in the Harvard College Alcohol Study (Weschler et al. 1999). Respondents rated the risk associated with each behavior (a "1" indicating "not risky at all" and a "7" indicating "extremely risky") for five consumption levels drunk at a single sitting. These levels included 1-2 beers, 3-4 beers, 5-7 beers, 8-9 beers, and 10 or more beers. Each respondent answered questions for all five behaviors at all five consumption levels.

Characterizations of Hypothetical Drinking Behaviors. The last set of measures asked respondents to rate several hypothetical drinking behaviors on a 7-point scale in terms of the severity of the drinking problem (1 indicated "abstention, alcohol not consumed," 4 indicated "moderate drinking," and 7 indicated "severe problem-drinking"). The measures varied the number of beers consumed on weekdays and on weekends. To minimize concerns with social desirability, respondents were reminded in the instructions that there were no right or wrong answers and that responses were anonymous and would remain confidential. Responses to these drinking behavior patterns were not linked to the hypothetical beer stimuli to which participants were exposed at the outset of the study. The drinking patterns included the following: (1) two or three beers each night during the week and ten or more on Friday and Saturday night; (2) ten or more beers consumed at one time a couple of times a week; (3) ten or more beers consumed at one sitting, about once a week, one or two per day the other days; (4) five or six beers every day of the week; (5) two or three beers on weeknights and five or six total on the weekend; (6) two or three beers every night throughout the week. The first four of these drinking patterns assessed substantially exceed the operational definition of binge drinking (i.e., defined as five drinks or more at a single sitting for men; four drinks or more for women). The final two measures do not directly exceed binge-drinking definitions, but they are greater than most recommendations for consumption by other health and advocacy groups (Dietary Guidelines for Americans 1999).

A subject debriefing occurred after the completion of the experiment. Subjects were presented a written debriefing statement explaining the purpose of the research and were able to ask questions of the experimenter. After discussing the experiment for several minutes, the normal class session began.

RESULTS

Effects on Drinking-Related Attitudes and Intentions

A multivariate analysis of variance (MANOVA) with follow-up univariate tests was used to test predictions in H1 concerning attitudes toward getting drunk, perceived social benefits associated with drinking beer, health benefits related to beer drinking, and product purchase intentions. Multivariate and univariate results are shown in Table 1, and means are shown in Table 2. Consistent with H1b and H1c, there are multivariate main effects of binge drinking status and country. The multivariate effect of warning message proposed in H1a is not significant. No multivariate interactions are significant.

Univariate follow-up tests indicate that means for these dependent variables all differ significantly across binge drinking status. Binge drinkers have much more favorable attitudes toward getting drunk (M = 4.3) than do non-binge drinkers (M = 2.5; F = 59.2; p < .01; eta-square = .19). Similarly, binge drinkers perceive stronger social benefits (M = 4.9; p < .01) associated with drinking than non-binge drinkers (M = 3.5; F 65.7; p < .01; eta-square .20). While means are below the scale midpoint for both groups, binge drinkers also view health benefits (M = 3.1) from drinking more favorably than non-bingers (M = 2.3; F = 19.6, p < .01; eta-square = .07), and bingers had stronger purchase intentions (M = 3.6) than non-bingers (M = 3.2; F = 4.5, p < .05; eta-square = .02).

Univariate follow-ups associated with respondents' country (Australia or U.S.) show significant effects for perceived health benefits (F = 4.8, p < .01) and purchase intentions (F = 24.9, p < .01). For both dependent variables, means for perceived health benefits (M = 3.0) and intentions (M = 3.9) for the Australian respondents were higher than those for the American respondents (M = 2.6 and 2.9, respectively). Means for the attitude toward getting drunk and social benefits were also somewhat higher for the Australians, but neither of these differences reached statistical significance (p < .11 for both). Thus, with the exception of the hypothesized effects of health warning message, Hypothesis 1 was generally supported.

Effects on Perceived Risk

Table 3 shows F-values for repeated measures analyses of the effect of the independent variables of consumption amount (the repeated measures factor), and binge drinking status, country of residence, and alcohol warning type. Cell means across the five levels are presented in Table 4. The effects of these factors were assessed for perceived risk associated with:

1) drinking when pregnant, 2) drinking and driving, 3) being arrested for driving under the influence or while impaired, 4) doing something that you would later regret after drinking, and 5) performing badly on a test or project because of drinking behavior (cf., Wechsler et al. 1997, 1999).

Hypotheses 2 and 3a predicted that the main effects of the independent variables would influence the levels of perceived risk associated with drinking-related behavior. As expected, with a few exceptions noted below, there was a main effect of type of health warning statement, binge-drinking status, country of residence, and consumption level on perceived risk for each of the five drinking-related behaviors. As shown by the means presented in Table 4, for each of the five drinking-related behaviors, less risk was perceived when the U.S. standard warning was presented (versus when alcohol was identified as a drug), by binge drinkers (versus non-binge drinkers), Australians (versus Americans), and at lower (versus higher) levels of consumption. Contrary to expectations, the risk of performing badly on a test was not influenced by country of residence nor was the risk associated with drinking and driving influenced by type of health-warning statement. However, these main effects need to be interpreted cautiously because of numerous two-way interactions between the independent variables.

Moderating Influence of Binge-drinking Status

The influence of consumption amounts on several of the perceived risk measures was moderated by binge-drinking status, as hypothesized in H3b. The strongest effect was found for the perceived risk associated with drinking and driving (F = 11.5, p < .01). As shown by the pattern of means presented in the top portion of Figure 1, binge drinkers perceive significantly less risk associated with driving after drinking than nonbingers at consumption amounts up to 7 beers. As shown in Table 4, at the 8-9 beer consumption level, risk means approach the maximum "ceiling" level for all groups except American bingers exposed to the U.S. warning (M = 6.0).

In addition to this above interaction for the drinking and driving behavior, the two-way drinking status by country interaction was significant (F = 6.4, p < .05), and the three-way consumption amount by drinking status by country interaction reveals a significant trend in the data (F = 2.45, p <.10). As shown by the plot of means in the center portion of Figure 1, the overall perceived risk associated with drinking and driving was quite similar for non-binge drinkers in both Australia and the U.S. However, overall, American binge drinkers perceived less risk (M = 5.3) than binge drinkers in Australia (M 5.9; p < .0 1); this plot again illustrates the difference between binge and non-binge drinkers.

The results also indicate that the effect of consumption amount on risks associated with performing badly on a test or project because of drinking was moderated by binge-drinking status (F = 5.9, p < .01). As shown in the lower portion of Figure 1, there are greater differences between the perceptions of binge and non-binge drinkers at lower levels of consumption. As consumption amounts increase, differences between the two groups, although they remain significant, decrease.

The results suggest an influence of the binge-drinking status by consumption amount interaction on the perceived risks associated with being arrested for DWI/DUI (F = 2.4, p < .10) and associated with doing something you would regret after drinking (F = 2.4, p < .10) although the findings do not reach significance. Similar to the other behaviors, binge drinkers perceived less risk than non-binge drinkers at lower consumption amounts but differences between the two groups decreased as consumption amounts increased.

Moderating Influence of Country

The influence of consumption amount on the perceived risks associated with drinking when pregnant (F = 14.1, p < .01) and with being arrested for DWI/DUI after drinking (F = 8.2, p < .01) was moderated by country, as predicted in H3b. As shown in Figure 2, the consumption amount by country interaction reveals greater differences between countries at lower levels of consumption; Australians perceive less risk than Americans. However, for both behaviors, differences between the countries decrease as consumption amounts increase. For the risk associated with drinking when pregnant, differences between countries remain significant at all consumption levels. For the perceived risk of DWI/DUI, follow-up tests show that the differences at the 1-2 and 3-4 beer levels between the two countries are significant (p < .01), but were no longer significant for amounts above the 3-4 beer level. The pattern of findings provides partial support for H3b.

Additional Significant Interactions Influencing Perceived Risk

Although not specifically hypothesized, several additional significant interactions provide further insight into some of the factors influencing the perceived risks associated with drinking-related behaviors. The perceived risk associated with doing something that you would later regret after drinking (F = 4.0, p < .05) and with performing poorly on a test or project because of drinking (F = 3.7, p < .10) were influenced by a significant interaction between binge-drinking status and warning type. The plots of the relevant means are shown in Figure 3. There is no difference in the effect of warning type for non-bingers, but for binge drinkers the "Alcohol is a Drug" warning leads to greater perceived risk of doing something later that would be regretted (M = 4.7) than the standard U.S. warning (M 4.1; p < .01). This interaction had a similar influence on the perceived risk of poor performance on a test or project because of drinking. While the "Alcohol is a Drug" message leads to greater risk perceptions overa ll, as indicated by the plot in the lower portion of Figure 3, there is little effect of the warning type for non-bingers. For binge drinkers, the "Alcohol is a Drug" warning leads to greater perceptions of risk (M = 5.5) than the standard U.S. warning (M = 4.9; p < .01).

Effects Related to Characterizations of Problem-Drinking

H4 pertains to perceptions of "problem-drinking" across various consumption levels. We were particularly interested in whether or not stating that "Alcohol is a Drug" as part of a warning would alter perceived problem-drinking levels. Multivariate and univariate effects for various drinking levels are shown in Table 5, and means are reported in Table 6. The drinking scenarios offered in the first four columns of Tables 5 and 6 all exceed the operational definition of binge drinking, and the other scenarios are above generally recommended consumption levels (Dietary Guidelines for Americans 1999).

Table 5 indicates a significant multivariate main effect of binge drinking status, and the univariate main effect means are significant or marginally significant for each of the drinking behavior patterns, offering support for H4a. While the level of the "drinking problem" was perceived as less severe by binge drinkers than by non-binge drinkers for each of the drinking patterns, these findings should be interpreted cautiously because of the interaction between binge drinking status and warning type (Wilks' lambda = .94, F = 2.79, p < .05). The univariate interaction is significant or marginally significant for four of the drinking behavior scenarios.

Plots for perceptions of the drinking behavior patterns are shown in Figure 4. (Plots for each of the dependent variables were similar.) Perceptions of non-bingers and bingers are very similar when the "Alcohol is a Drug" warning is used, but all perceptions differ significantly between binge and non-binge drinkers when the standard U. S. warning is used (t-values range from 2.61 to 4.60, p < .01 for all). These findings support the pattern predicted in H4b. For the behavior of drinking 2-3 beers per day, for binge drinkers, perceptions of the behavior as a more severe drinking problem are greater when the "Alcohol is a Drug" warning is presented (t = 2.l9, p < .05) than when the standard U. S. warning is presented. For the behavior of drinking 10 or more beers once a week, with 1-2 on the other days, the difference between warnings for binge drinkers does not reach statistical significance. However, as shown by the pattern of means in Table 6, all ten cells for binge drinkers shown the "Alcohol is a Drug" wa rning resulted in higher directional means for perceptions of problem drinking as compared to the binge drinkers exposed to the standard U.S. warning.

DISCUSSION

Alcohol-related traffic accidents have been cited as the leading cause of death for people between 15 and 24 years of age in the United States (National Committee for Injury Prevention and Control 1989; U.S. Dept. of Education 1994), and, as previously mentioned, college presidents identify binge drinking as the most important problem on campuses today. Given compelling evidence of the magnitude of the problem, the purpose of this study was to examine effects on various drinking attitudes and intentions, perceived risk, and perceptions of consumption patterns for two beverage warnings across consumers classified as binge and non-binge drinkers from two different countries. Findings offer mixed results for the effect of the type of warning ("Alcohol is a Drug" versus the standard warning used in the U.S.). The warning type had no significant effect on social or health benefits or other attitudinal variables, but it had diverse effects on the perceived risk measures and perceptions of various drinking behaviors . While the standard U.S. warning explicitly mentions the risks of drinking when pregnant and drinking when driving a car, greater risk perceptions were associated with the "Alcohol is a Drug" warning for both drinking when pregnant and risk of a DWI/DUI citation. This finding is particularly interesting for the risk of drinking when pregnant, given that this behavior is the first risk noted and is explicitly stated in the U.S. warning. Similarly, the "Alcohol is a Drug" warning led to greater perceptions of performing poorly on a test or project. For two of the risk perception measures, there was some evidence that the type of warning had a greater effect on binge drinkers. For binge drinkers, use of the standard U.S. warning resulted in lower risk perceptions than the "Alcohol is a Drug" warning (see Figure 3).

Similar results were found for effects of the warning type across binge and non-binge drinkers on the perceptions of various drinking behaviors. After being exposed to the "Alcohol is a Drug" warning, binge drinkers' perceptions of the drinking patterns are more strongly impacted than perceptions of non-bingers. Non-binge drinkers tend to view the drinking patterns as more problematic than do binge drinkers when the standard warning is used; however, there is no difference between bingers and non-bingers when they have been exposed to the "Alcohol is a Drug" warning.

This pattern of results is consistent with an associative memory model perspective (Anderson 1983; Collins and Loftus 1975). Positioning alcoholic beverages as "a drug" is more likely to create cognitive links to more negative concepts (e.g., addiction, hard drugs like cocaine and amphetamines for which risks are more well accepted) that influence risk and consumption perceptions. While frequent and heavier drinkers have been found to discount the standard U.S. warning (Andrews and Netemeyer 1996), these results suggest that discounting of risks by binge drinkers may be a less likely outcome for the "Alcohol is a Drug" warning.

In general, our findings reinforce suggestions that new, potentially stronger warnings such as "Alcohol is a Drug" should be considered as one possible replacement for the current U.S. warning in a system in which different warnings are rotated (Hilton 1993; Andrews and Netemeyer 1996). However, it should be noted that after repeated exposures this warning message might wear out and thus lose its effectiveness. Although U.S. drinkers have been repeatedly exposed to both the container message and messages from other sources regarding risks associated with drinking and driving, at lower levels of consumption (e.g., 1-2 and 3-4 beers), binge drinkers in the U.S. perceived less risk than either binge drinkers in Australia or non-bingers in either country (Christie et al. 2001). The pattern of lower risk perception means found for the binge drinkers in this study is consistent with prior research that shows that more frequent and heavier drinkers tend to discount the alcohol warning information (Andrews et al. 199 1; Andrews and Netemeyer 1996).

Study results also reinforced much of the prior large cross-sectional survey findings that show much more favorable attitudes toward getting drunk and greater perceived social and health benefits of beer consumption for binge drinkers (Wechsler et al. 1994, 1999). In addition, binge drinkers generally perceived less of a problem with the drinking behavior patterns assessed, and perceived less risk associated with more moderate consumption levels. Thus, binge drinkers exhibited the feelings of personal immunity from harm that consumers who consider themselves more careful or skilled with a product than others often exhibit. Our findings, coupled with prior results for the binge segment, suggest that a more targeted communication approach aimed at changing attitudes and ultimately behavior of the binge-drinking segment may be most useful in impacting this group. While this is clearly a daunting task, the findings in this study that show that certain warning messages can have stronger effects for the binge segme nt (see Figures 3 and 4) and suggest that certain types of communication may be more effective for this more vulnerable group.

To our knowledge, there has been little prior experimental research examining warnings and binge drinking in a cross-cultural study. While general support was found for expectations of somewhat more positive attitudes toward benefits of drinking for the Australians, there were not major differences in warning effects or drinking behavior pattern perceptions across subjects from the two countries. However, the risks associated with alcohol consumption were often viewed differently across the two countries. The largest difference (eta-square = .11) was the risk associated with drinking when pregnant; Australians at all consumption levels perceived less risk of drinking when pregnant, and the differences were larger at lower consumption levels. These results suggest that the message about drinking when pregnant on the bottle labels and elsewhere have created high levels of awareness in the U.S., and they indicate additional efforts at communicating this risk seem needed in Australia.

LIMITATIONS AND FUTURE RESEARCH

This study examined two types of alcoholic beverage warnings, one currently used in the U.S. and one that was considered for possible use in Australia. Results are limited to these specific warnings, and wording modifications to the warning used may affect findings. Also, similar to other experiments in which various alcoholic beverage warnings have been tested (Andrews et al. 1990, 1991), there was forced exposure to the beer stimuli and warning information in a classroom setting and on a single occasion. These conditions differ considerably from actual marketplace exposure to warnings, and results from this study may not generalize to other exposure conditions. The subjects sampled, some of whom may have been abstainers, were undergraduate students from two universities, one located in Australia and one in the U.S. Student consumers are a highly relevant segment for studies involving alcohol warnings and binge drinking and are frequently used in experimental studies on warnings (Andrews and Netemeyer 1996), but results should not be generalized to other segments of the population. Future research may address effects for other segments; examine potential differences of warnings for males and females; assess findings for very infrequent drinkers, moderate consumers, and abstainers; and investigate warning effects across different contexts.

Other future studies also appear warranted. For example, the rationale for effects of an "Alcohol is a Drug" warning was based around an associative network of memory in which more negative consequences and risks would be evoked by presenting alcohol as a drug. Studies that analyze consumers' thoughts while they read and consider an alcohol health-warning statement (i.e., their cognitive responses) would be particularly beneficial. There is some concern that presenting alcohol as a drug may make it be seen as more of a "forbidden fruit" for certain segments of the population and may increase its appeal for at least some consumers. Also, past research has found that support arguments mediated about three-quarters of the effect of warning types on label attitude, and future studies may address such effects across binge and non-binge drinkers. Thus additional research such as this might provide new insight regarding how different types of cognitive processes have differing effects on consumers' responses to health-warning statements.

There is mounting evidence that any incremental benefits of continued use of the current U.S. warning are declining, and that use of factual warnings rotated across labels has several desirable characteristics from a public policy perspective (Andrews and Netemeyer 1996; Mazis 1990). This suggests that reviews and future research address other warning possibilities (Slater et al. 1995, 1998), and perhaps that other formats and specifications be considered. Research examining effects related to novelty of warnings and potential "wear-out" of warnings used for many years for different segments of the target market are needed.

Lastly, experimental studies of the binge-drinking behaviors across different countries are rare, despite a burgeoning interest in cross-national studies with policy implications (Bamossy, Belk, and Costa 1997). The two countries used in this study were ideally suited for data collection because Australia has not required any warning while a warning had been used in the U.S. since 1989. However, these are both western countries characterized by positive attitudes toward drinking behaviors and alcoholic beverages, especially beer, the product used in this study. Future research may extend findings to other countries where attitudes towards alcohol are less positive, there is greater collectivism rather than individualism, or where other types of alcohol-related warnings had been used. Unfortunately, binge-drinking and other forms of alcohol abuse are not unique to the United States and Australia. Excessive alcohol consumption is associated with a wide variety of negative outcomes such as property damage, acade mic failure, alcohol poisoning, sexually transmitted diseases, violence, personal injury, illness, and death in societies across the globe. A better understanding of the factors influencing the effectiveness of health warnings will help reduce the high incidence of binge-drinking behaviors and, in turn, will have a positive impact on the health, safety, and welfare of many people around the world.

[FIGURE 1 OMITTED]

[FIGURE 2 OMITTED]

[FIGURE 3 OMITTED]

[FIGURE 4 OMITTED]

TABLE 1

Multivariate and Univariate Results for Drinking-Related Attitudes and
Purchase Intentions

                             Multivariate Results       Univariate
                                                         F Values


Independent
Variable                Wilks' [lambda]     F Value     df

Main Effects
  Drinking Status (DS)        .72            24.5 *   (1,260)
  Culture (C)                 .91             6.7 *   (1,260)
  Warning (W)                 .99             0.7     (1,260)
Interactions
  DS X C                      .97             1.7     (1,260)
  DS X W                      .99             0.3     (1,260)
  C X W                       .98             1.5     (1,260)
  DS X W X C                  .99             0.8     (1,260)

                                   Univariate F Values
                        Attitude
                         Toward
Independent             Getting     Social    Health   Purchase
Variable                 Drunk     Benefits  Benefits  Intention

Main Effects
  Drinking Status (DS)   59.2 *     65.7 *    19.6 *      4.5 **
  Culture (C)             2.7        2.7       4.8 *     24.9 *
  Warning (W)             0.4        0.1       1.1        2.2
Interactions
  DS X C                  1.3        4.0 **    0.5        1.7
  DS X W                  0.2        0.0       0.9        0.2
  C X W                   1.6        2.0       0.0        0.1
  DS X W X C              0.0        0.0       0.0        3.1

* p < .01

** p < .05

TABLE 2

Means for Drinking-Related Attitudes and Purchase Intentions

                              Warning ="Alcohol is a Drug"

                    Attitude
                     Toward
Nation/Drinker      Getting       Social         Health   Purchase
Conditions           Drunk       Benefits       Benefits  Intention

Australia
 Non-Binge Drinker    2.95         3.66           2.49      3.56
 Binge Drinker        4.39         4.74           3.20      3.84

United States
 Non-Binge Drinker    2.01         3.29           1.86      2.47
 Binge Drinker        3.96         5.04           3.08      3.14

                               Warning = U.S. Standard

                    Attitude
                     Toward
Nation/Drinker      Getting       Social      Health   Purchase
Conditions           Drunk       Benefits    Benefits  Intention

Australia
 Non-Binge Drinker    2.69         4.00        2.84      3.63
 Binge Drinker        4.33         5.00        3.26      4.42

United States
 Non-Binge Drinker    2.33         3.12        2.22      3.11
 Binge Drinker        4.52         4.83        3.06      2.96

TABLE 3

F-Values Associated With Perception of Perceived Risk Across Varying
Consumption Levels


                          Drinking                            Doing
                            when     Driving                Something
Independent Variables     Pregnant    a Car     DWI/DUI     You Regret

Main Effects:
 Consumption Amount (CA)    97.2 *    351.1 *    152.7 *      237.5 *
 Drinking Status (DS)        5.1 *     35.0 *      9.5 *       59.4 *
 Culture (C)                32.4 *      5.2 **     9.6 *        4.2 **
 Warning (W)                 3.9 **     1.7        4.3 **       3.3 ***

Interactions (a):
 CA X DS                     0.2       11.5 *      2.4 ***      2.7 ***
 CA X C                     14.1 *      0.7        8.2 *        1.3
 CA X W                      0.5        1.2        0.3          0.1
 DS X C                      1.2        6.4 **     1.1          0.3
 DS X W                      0.3        0.2        2.2          4.0 **
 C X W                       1.3        0.3        0.0          0.5

                          Performing
                          Poorly on
                           Test or
Independent Variables      Project

Main Effects:
 Consumption Amount (CA)    232.0 *
 Drinking Status (DS)        29.9 *
 Culture (C)                  0.6
 Warning (W)                  4.1 **

Interactions (a):
 CA X DS                      5.9 *
 CA X C                       1.4
 CA X W                       1.9
 DS X C                       0.4
 DS X W                       3.7 ***
 C X W                        0.0

* p < .01.

** p < .05.

*** p < .10.

(a)All three- and four-way interactions are nonsignificant at p < .05.

TABLE 4

Means for Risk Perception Dependent Variables

                                 Warning = "Alcohol is a Drug"

Nation/Binge-Drinker
Conditions                   1-2 Beers  3-4 Beers    5-7 Beers

Pregnancy
  Australians
    Non-Binge Drinker          5.39       6.11         6.72
    Binge Drinker              5.33       5.70         6.38
  U.S.A.
    Non-Binge Drinker          6.49       6.72         6.92
    Binge Drinker              6.33       6.49         6.82

Driving a Car
  Australians
    Non-Binge Drinker          4.83       6.08         6.89
    Binge Drinker              3.84       5.36         6.40
  U.S.A.
    Non-Binge Drinker          4.92       6.00         6.71
    Binge Drinker              3.27       4.71         6.04

DUI/DWI
  Australians
    Non-Binge Drinker          4.61       5.56         6.42
    Binge Drinker              4.52       5.36         6.40
  U.S.A.
    Non-Binge Drinker          5.95       6.42         6.42
    Binge Drinker              5.09       5.82         5.82

Doing Something You Later
Regret
  Australians
    Non-Binge Drinker          3.89       5.00         5.94
    Binge Drinker              3.16       3.68         4.68
  U.S.A.
    Non-Binge Drinker          4.87       5.61         6.03
    Binge Drinker              3.53       4.20         5.40

Performing Poorly on a Test
or Project
  Australians
    Non-Binge Drinker          4.77       5.34         6.31
    Binge Drinker              3.96       4.76         6.04
  U.S.A.
    Non-Binge Drinker          5.26       5.89         6.42
    Binge Drinker              4.11       4.87         5.96

                             Warning = "Alcohol is a     Warnin g = U.S.
                                      Drug"                  Standard

Nation/Binge-Drinker                    10 or More
Conditions                   8-9 Beers    Beers     1-2 Beers  3-4 Beers

Pregnancy
  Australians
    Non-Binge Drinker          6.89        7.00       5.12       5.77
    Binge Drinker              6.58        6.75       5.00       5.64
  U.S.A.
    Non-Binge Drinker          6.97        7.00       6.47       6.64
    Binge Drinker              7.00        7.00       5.92       6.56

Driving a Car
  Australians
    Non-Binge Drinker          6.94        7.00       4.39       5.87
    Binge Drinker              6.76        6.96       4.36       5.43
  U.S.A.
    Non-Binge Drinker          6.87        6.97       4.79       6.06
    Binge Drinker              6.67        6.89       3.33       4.37

DUI/DWI
  Australians
    Non-Binge Drinker          6.72        6.83       4.67       5.60
    Binge Drinker              6.76        6.84       3.83       5.00
  U.S.A.
    Non-Binge Drinker          6.92        7.00       5.60       6.26
    Binge Drinker              6.80        6.96       4.81       5.44

Doing Something You Later
Regret
  Australians
    Non-Binge Drinker          6.42        6.81       4.29       5.10
    Binge Drinker              5.04        5.48       2.57       3.21
  U.S.A.
    Non-Binge Drinker          6.32        6.63       4.70       5.43
    Binge Drinker              5.93        6.33       2.93       3.37

Performing Poorly on a Test
or Project
  Australians
    Non-Binge Drinker          6.60        6.69       4.64       5.48
    Binge Drinker              6.28        6.48       3.43       4.00
  U.S.A.
    Non-Binge Drinker          6.60        6.84       5.01       5.62
    Binge Drinker              6.24        6.51       3.67       4.11

                                  Warnin g = U.S. Standard

Nation/Binge-Drinker                                10 or More
Conditions                   5-7 Beers   8-9 Beers    Beers

Pregnancy
  Australians
    Non-Binge Drinker          6.38        6.68        6.90
    Binge Drinker              5.93        5.92        6.14
  U.S.A.
    Non-Binge Drinker          6.83        6.89        7.00
    Binge Drinker              6.89        6.92        6.74

Driving a Car
  Australians
    Non-Binge Drinker          6.68        6.87        7.0
    Binge Drinker              6.00        6.57        7.0
  U.S.A.
    Non-Binge Drinker          6.68        6.89        6.96
    Binge Drinker              5.37        6.00        6.51

DUI/DWI
  Australians
    Non-Binge Drinker          6.36        6.53        6.77
    Binge Drinker              5.75        6.25        6.67
  U.S.A.
    Non-Binge Drinker          6.74        6.91        6.98
    Binge Drinker              5.89        6.30        6.44

Doing Something You Later
Regret
  Australians
    Non-Binge Drinker          5.90        6.42        6.65
    Binge Drinker              4.14        4.43        5.36
  U.S.A.
    Non-Binge Drinker          6.23        6.51        6.72
    Binge Drinker              4.26        5.15        5.22

Performing Poorly on a Test
or Project
  Australians
    Non-Binge Drinker          6.26        6.65        6.84
    Binge Drinker              5.07        5.64        6.36
  U.S.A.
    Non-Binge Drinker          6.36        6.74        6.94
    Binge Drinker              5.00        5.67        6.11

TABLE 5

Multivariate and Univariate Results for Perceptions of Drinking Behavior
Patterns

                            Multivariate Results        Univariate F
                                                         Values (a)

                                                    10 or More on
                                                     Both Friday
                                                     and Saturday
Independent Variables    Wilks' [lambda]  F Value   2-3 Other Days

Main Effects:
   Drinking Status (DS)        .95         2.23 **       3.4 ***
   Culture (C)                 .96         1.75          0.1
   Warning (W)                 .98         0.98          0.2

Interactions:
   DS X C                      .99         0.40          0.0
   DS X W                      .94         2.79 **       3.0 ***
   C X W                       .95         2.39 **       0.2
   DS X C X W                  .98         0.66          0.0

                                 Univariate F Values (a)

                                       10 or More
                         10 or More   Once a Week,  5-6 Every
                          Twice a      1-2 Other    Day of the
Independent Variables       Week          Days         Week

Main Effects:
   Drinking Status (DS)     3.8 ***       5.8 **       3.8 ***
   Culture (C)              1.3           0.1          1.2
   Warning (W)              1.9           0.0          0.5

Interactions:
   DS X C                   0.0           0.5          0.0
   DS X W                   0.1           5.0 **       1.1
   C X W                    0.1           0.1          0.6
   DS X C X W               0.5           0.0          0.5

                          Univariate F Values
                                  (a)

                          5-6 on
                         Weekends,  2-3 Each
                         2-3 Other   Day of
Independent Variables      Days     the Week

Main Effects:
   Drinking Status (DS)    11.9 *     3.2 ***
   Culture (C)              0.4       0.8
   Warning (W)              0.0       0.9

Interactions:
   DS X C                   0.0       0.2
   DS X W                   6.5 **    5.2 **
   C X W                    0.1       2.7
   DS X C X W               0.0       0.1

* p < .01

** p < .05

*** p < .10

(a)Degrees of freedom = (1.257) for all univariate tests.

TABLE 6

Means for Drinking Behavior Patterns

                                10 or More
                                 on Both
                                Friday and              10 or More
                                Saturday,   10 or More    Once a
                                2-3 Other    Twice a    Week, 1-2
                                   Days        Week     Other Days

Warning = "Alcohol is a Drug":
 Australia
  Non-Binge Drinker                5.50        5.82        4.88
  Binge Drinker                    5.54        5.21        4.75
 United States
  Non-Binge Drinker                5.45        5.40        4.87
  Binge Drinker                    5.33        5.04        4.93

Warning = U.S. Standard:
 Austraila
  Non-Binge Drinker                5.71        5.26        5.39
  Binge Drinker                    5.00        5.07        4.36
 United States
  Non-Binge Drinker                5.75        5.24        5.18
  Binge Drinker                    5.00        4.71        4.46



                                            5-6 on
                                5-6 Every  Weekends,  2-3 Each
                                 Day of    2-3 Other   Day of
                                the Week     Days     the Week

Warning = "Alcohol is a Drug":
 Australia
  Non-Binge Drinker               5.47       4.29       4.62
  Binge Drinker                   5.21       4.17       4.58
 United States
  Non-Binge Drinker               5.29       4.26       4.37
  Binge Drinker                   5.24       4.09       4.58

Warning = U.S. Standard:
 Austraila
  Non-Binge Drinker               5.54       4.81       4.52
  Binge Drinker                   5.14       3.79       3.79
 United States
  Non-Binge Drinker               5.35       4.65       4.94
  Binge Drinker                   4.57       3.64       4.25

ENDNOTES

(1.) Binge drinking is defined as five or more drinks in a single time period for a male and four or more drinks for a female in many recent studies (Wechsler et al. 1994, 1999).

(2.) In addition to these behaviors related to driving and drinking when pregnant, we examine the behaviors of performing poorly on a test or project and doing something you later regretted. These behaviors seemed relevant for the student sample and are all drawn from prior research on alcohol consumption and hinge drinking (Wechsler et al. 1994,1999).

(3.) Specific measures are available upon request from the first author.

REFERENCES

Azjen, Icek and Martin Fishbein. 1980. Understanding Attitudes and Predicting Social Behavior. Englewood Cliffs, N.J,: Prentice Hall.

Anderson, John R. 1983. The Architecture of Cognition. Cambridge, MA: Harvard University Press.

Andrews, J. Craig and Richard G. Netemeyer. 1996. Alcohol Warning Label Effects: Socialization, Addiction, and Public Policy Issues. In Marketing and Consumer Research in the Public interest, edited by Ronald Paul Hill. Thousand Oaks, CA: Sage Publications, 153-175.

Andrews, J. Craig, Richard G. Netemeyer, and Srinivas Durvasula. 1990. Believability And Attitudes Toward Alcohol Warning Label Information: The Role of Persuasive Marketing Communications Theory. Journal of Public Policy & Marketing, 9(1): 1-15.

Andrews, J. Craig, Richard G. Netemeyer, and Srinivas Durvasula. 1991. Effects of Consumption Frequency on Believability and Attitudes Toward Alcohol Warning Labels. The Journal of Consumer Affairs, 25(2): 323-338.

Andrews, J. Craig, Richard G. Netemeyer, and Srinivas Durvasula. 1993. The Role of Cognitive Responses as Mediators of Alcohol Warning Label Effects. Journal of Public Policy & Marketing, 12(1): 57-68.

Bamossy, Gray J., Russell Belk, and Janeen Arnold Costa. 1997. International Issues in Law and Public Policy: Special Editors Note, Journal of Public Policy & Marketing, 16(1): 1.

Bushman, Brad J. and Angela Stack. 1996. Forbidden Fruit Versus Tainted Fruit: Effects of Warning Labels on Attraction to Television Violence. Journal of Experimental Psychology, Applied 2(3): 207-226.

CASA Commission on Substance Abuse at Colleges and Universities. 1994. Rethinking Rites of Passage: Substance Abuse on America's Campuses--A Report by the Commission on Substance Abuse at Colleges and Universities, New York: Center on Addiction and Substance Abuse at Columbia University.

Center for Science in the Public Interest. 2000. Fact Sheet: Binge Drinking on College Campuses, Booze News, http://www.cspinet.org.

Christie, Jennifer, Dan Fisher, John C. Kozup, Scott Smith, Scot Burton, and Elizabeth H. Creyer. 2001. The Effects of Bar Sponsored Alcohol Beverage Promotions Across Binge and Non-Binge Drinkers. Journal of Public Policy & Marketing, 20(2): 240-253.

Collins, Allan M. and Elizabeth F. Loftus. 1975. A Spreading Activation Theory of Semantic Processing. Psychological Reports, 82 (6): 407-428.

Dietary Guidelines for Americans. 1999. Washington D.C., U.S. Department of Agriculture; U.S. Department of Health and Human Services.

Greenfield, Thomas K., Karen L. Graves, and Lee A. Kaskutas. 1999. Long-Term Effects of Alcohol Warning Labels: Findings from a Comparison of the United States and Ontario, Canada, Psychology and Marketing, 16(3): 261-282.

Hankin, Janet R., James J. Sloan, and Robert Sokol. 1998. The Modest Impact of the Alcohol Beverage Warning Label on Drinking During Pregnancy Among a Sample of African-American Women. Journal of Public Policy and Marketing, 12 (1): 61-69.

Hanson, David J. and Ruth C. Engs. 1992. College Students' Drinking Problems: A National Study. Psychological Reports, 71: 39-42.

Hilton, Michael E. 1993. An Overview of Recent Findings on Alcohol Beverage Warning Labels. Journal of Public Policy & Marketing, 12 (1): 1-9.

Johnson, Vernon E. 1980. I'll Quit Tomorrow. San Francisco: Harper & Row,

Kerr, William C., Kaye Middleton Fillmore, and Paul Marvy. 2000. Beverage-Specific Alcohol Consumption and Cirrhosis Mortality in a Group of Englihs-speaking Beer-Drinking Countries. Addiction, 96(3): 339-346.

Mazis, Michael B. 1990. The Marketing of Alcohol: A Spirited Debate. In Marketing and Advertising Regulation, edited by P. Murphy & W. Wilkie. Notre Dame, IN: University of Notre Dame Press, 221-233.

National Committee for Injury Prevention and Control. 1989. Injury Prevention: Meeting the Challenge. New York: Oxford University Press.

National Rural Health Alliance. 1998. Drugs and Alcohol in Rural Australia. Rural Health Information Paper No.4 (August).

Pittel, Stephen. 1985. Drug Abuse Treatment Modalities: An Overview. In Addictive Behavior: Drug and Alcohol Abuse. Englewood, CO: Morton: 327-333.

Presley, Cheryl A., Philip W. Meilman, and Rob Lyeria. 1993. Alcohol and Drugs on American College Campuses: Use, Consequence, and Perceptions of the Campus Environment, Volume 1: 1989-1991. Carbondale, IL: Core Institute.

Reuters News Service. 1994. Australians May Be Biggest Binge Drinkers, Says Expert. Food and Drink Daily (June 28):3.

Slater, Michael D., David Karan, Donna Rouner, Kevin Murphy, and Frederick Beauvais. 1998. Developing and Assessing Alcohol Warning Content: Responses to Quantitative Information and Behavioral Recommendations in Warnings with Television Beer Advertisements. Journal of Public Policy & Marketing, 17 (1):48-60.

Slovic, Paul, Baruch Fischoff, and Sarah Lichtenstein 1980. "Facts & Fears: Understanding Peceived Risk. In Societal Risk Assessment: How Safe Is Safe Enough? edited by R.C. Schwing and W.A. Albers, Jr. NY: Plenum Press: 181-214.

Svenson, Ola, Baruch Fischoff, and Daniel Macgregor. 1985. Perceived Driving Safety and Seatbelt Usage. Accident Analysis and Prevention, 17:119-133.

Treise, Debbie, Joyce M. Wolburg, and Cele C. Otnes. 1999. Understanding the "Social Gifts" of Drinking Rituals: An Alternative Framework for PSA Developers. Journal of Advertising, 28 (2):17-31.

U.S. Dept. of Education. 1994. Youth and Alcohol: Selected Reports to the Surgeon General. Washington, D.C.

Wechsler, Henry A. and Nancy E. Issac. 1992. Binge Drinkers at Massachusetts Colleges. Journal of the American Medical Association, 267 (21):2929-2931.

Wechsler, Henry A., A. Davenport, G. Dowdall, B. Moeykens, and S. Castillo. 1994. Health and Behavioral Consequences of Binge Drinking in College: A National Survey of Students at 140 Campuses. Journal of the American Medical Association, 272 (21):1672-77.

Wechsler, Henry A., B. Molnar, A. Davenport, and J. Baer. 1999. College Alcohol Use: A Full or Empty Glass? Journal of American Collegiate Health, 47 (6):247-252.

Wechsler, Henry A. and S. B. Austin. 1998. Binge Drinking: The Five/Four Measure Journal of Studies on Alcohol, 59 (l):122-124.

Wechsler, Henry A., Jae Eun Lee, Meichum Kuo, and Hang Lee. 2000. College Binge Drinking in the 1990's: A Continuing Problem. Journal of American Collegiate Health, 48 (5):199-210.

Elizabeth H. Creyer is an Associate Professor and Scot Burton is a Professor in the Marketing and Transportation Department, Sam M. Walton College of Business Administration, University of Arkansas, Fayetteville. John C. Kozup is an Assistant Professor at Villanova University.