The effect of cancer warning statements on alcohol consumption intentions

Abstract

In response to increasing calls to introduce warning labels on alcoholic beverages, this study investigated the potential effectiveness of alcohol warning statements designed to increase awareness of the alcohol–cancer link. A national online survey was administered to a diverse sample of Australian adult drinkers (n = 1,680). Along with attitudinal, intentions and demographic items, the survey included an online simulation that exposed respondents to one of six cancer warning statements delivered across a range of situational contexts. Half of the statements made general reference to cancer and half mentioned specific forms of cancer. Respondents reported on the believability, convincingness and personal relevance of the warning statements. Pre- and post-exposure data were captured relating to respondents’ alcohol consumption intentions. Of the six statements tested, Alcohol increases your risk of bowel cancer produced the highest scores across all outcome measures. All statements produced favorable changes in alcohol consumption intentions, including among high-risk drinkers. There is thus the potential for these and similar statements to be used as a suite of rotating warning messages located on alcoholic beverage labels and applied in various public education contexts.

Introduction

The World Health Organization (WHO) [1] recognizes alcohol as one of the top three causes of poor health. In Australia, alcohol-related harms are estimated to cost almost $AU30 billion annually [2]. In alignment with the WHO’s recommendation to include health warnings on alcohol packaging, reviews of Australian health policies have called for warning statements to be included on alcohol products to overcome knowledge deficits relating to the health risks associated with alcohol consumption and to assist drinkers make informed decisions [3, 4]. There is strong community support for this approach [5]. However, despite considerable work in this domain in the context of tobacco control, it has been noted that there is very little evidence about the types of messages that could be effective for alcohol [5–7]. Most of the limited research in this area has focused on youth [8–10], and as such there is a lack of evidence to guide the introduction of warning statements that have the potential to be effective at a population level. There is also a lack of evidence relating to whether such warnings would be effective in contexts other than product packages, such as in mass media campaigns, news stories and information provided by health practitioners.

A growing evidence base implicates alcohol consumption at even low levels as a risk factor for a range of cancers [11, 12]. While many adults are aware of the association between alcohol and illnesses such as heart disease and liver cirrhosis [5], few understand the increased risk of cancer resulting from alcohol consumption [13]. This situation has resulted in calls for cancer warning statements to be placed on alcoholic beverages to (i) increase awareness of the alcohol–cancer link [14], (ii) potentially provide strong motivation to reduce consumption given current high levels of fear of cancer in the community [13] and (iii) constitute a modest but important countermeasure to the alcohol industry’s enormous investment in alcohol promotion [15].

In response to these calls, the present study investigated the extent to which a series of cancer warning statements was perceived to be believable, convincing and personally relevant to Australian drinkers. These message attributes were selected based on previous tobacco control research using these variables [16–18], and evidence that perceptions of messages reflect their effectiveness in terms of intended and actual behavior change [19, 20]. In combination, these attributes provide an indication of the extent to which drinkers may consider the information generally trustworthy and relevant to their specific circumstances. In addition, they reflect aspects of the health belief model that highlights the need for individuals to understand the existence of a threat to health and to perceive themselves to be personally vulnerable to this threat [21].

The effects of the messages on drinkers’ alcohol consumption intentions were assessed via an online simulation that imitated various scenarios in which drinkers could be feasibly exposed to cancer warning statements relating to alcohol. Such an approach recognizes that warning statements on products cannot be expected to produce substantial behavioral changes unless accompanied by other forms of information dissemination [4, 22].

Methods

Prior to this study, a series of focus groups (six groups, n = 48) and a large-scale online survey of adult Australian drinkers (n = 2,168) were used to develop a series of 11 cancer warning statements that are consistent with relevant communications and health behavior theories, considered acceptable by drinkers and appropriate for inclusion on alcoholic beverage packages [23]. The resulting statements varied according to the type of cancer mentioned (general versus specific), causation wording (‘can cause cancer’ versus ‘increases risk of cancer’), message framing (positive versus negative) and inclusion of the term ‘Warning’ (or not).

On the basis of the results, a subset of six statements was selected for further testing across a broader range of information provision contexts. A second online survey was subsequently administered via the same large web panel provider (PureProfile) to a separate sample of Australian adults. Ethics clearance for the study was obtained from the University of Western Australia Human Research Ethics Committee. As per the first survey, a large web panel provider was contracted to provide the sample. PureProfile has a panel of 350 000 Australians with diverse geographic and socioeconomic profiles who receive small financial incentives for participating in surveys. Multiple recruitment strategies have been used to establish the PureProfile panel, including referrals, internet and radio advertising and publicity. Potential respondents could either respond to an email distributed to panel members or they could access the survey link from the web panel provider’s website. Eligibility criteria were non-inclusion in the first survey, age (18–65 years) and the consumption of alcohol on at least two days per month (assessed by asking ‘In the last 12 months, how often did you have an alcoholic drink of any kind?’). Quotas were used to achieve an even gender split across three main age categories (<31 years, 31–45 years, 46–65 years). Table I provides the resulting sample profile (n = 1,680).

Table I.

Sample profile (n = 1,680)

Gender	%
Male	50.1
Female	49.9
Age
<31 years	49.5
31–45 years	25.2
46–65 years	25.3
Tertiary education (%)
Yes	46.1
No	53.9
SESˆ(%)
Low	28.4
Mid	43.3
High	28.4
Mean standard drinks per week	9.27 (SD = 13.60)
Long-term risk (%)
Low	81.6
High*	18.4
Short-term risk (%)
Low	26.5
High*	73.5

^ˆSocioeconomic status as per the Australian Bureau of Statistics’ Socio-Economic Indexes for Areas (SEIFA) classification [24].

^*High long-term risk = average of more than two standard drinks per day; High short-term risk = more than four standard drinks in a single sitting [25].

The alcohol consumption measures were those used in national alcohol intake surveys [25, 26] and involved respondents reporting the frequency with which they consumed alcohol in the previous 12 months, the number of standard drinks consumed on a usual drinking occasion, and the largest number of standard drinks consumed during one sitting in the previous 12 months. The first two items were used to create an average number of drinks consumed per day.

In addition to the alcohol consumption and attitudinal items used in the first survey, the second survey included behavioral intention measures to investigate the potential for the statements to encourage drinkers to modify their drinking practices. Future consumption intentions were measured by asking respondents: (i) the extent to which they believed they should reduce the amount of alcohol they consume (5-point scale: (1) not at all to (5) to a very great extent) (based on [27]), (ii) the extent to which they expected that they will actually reduce the amount of alcohol they consume (5-point scale: (1) not at all to (5) to a very great extent) (based on [27]) and (iii) their intention to consume five or more drinks in a single sitting within the following two weeks (5-point scale of (1) definitely intend not to (5) definitely intend to) (as per [28]). These questions were asked prior to and after exposure to one of the cancer warning statements via the simulation.

As part of the survey, respondents entered the online simulation that involved navigating through three locations: a home living room, a doctor’s office and a road-side bus stop. Respondents were randomly allocated to commence their simulation experience in either the living room or the doctor’s office, with the bus stop encountered when moving between these two locations. While within the simulation, each respondent was randomly exposed to one of the six statements that was presented in five different forms: an advertisement in a newspaper, a comment made by a child about information learned during a health class at school, a warning on an alcohol product (the type of alcoholic beverage was determined by their previously stated preference), a billboard at the bus stop and a statement made by a doctor. Each respondent therefore saw the same message repeated five times in differing contexts within the simulation, mimicking a situation in which alcohol warning statements are disseminated via a comprehensive community education program.

Upon leaving the simulation, respondents answered questions relating to their perceptions of the believability, convincingness and personal relevance of the warning statement to which they were exposed. Perceived believability and convincingness (How believable/unbelievable did you find this message? and How convincing/unconvincing did you find this message?) were measured on 5-point scales: 1 (Not at all believable/convincing) to 5 (Very believable/convincing). Personal relevance (How much do you feel this message applies to you?) was also measured on a 5-point scale: 1 (It does not apply to me at all) to 5 (It directly applies to me).

Analysis

The attitude outcome variables of believability, convincingness and personal relevance were then examined across all statements and by each statement individually. A composite score comprising all three attitude variables (‘attitudinal composite’) was derived by calculating the grand mean of all attitudinal outcome variables. Changes in drinking intentions after exposure to the statements were examined across all statements and by each statement individually. A composite score comprising all three intention variables (‘intention composite’) was also derived by calculating the grand mean of all intention outcome variables.

To examine the significance of pre- to post-exposure changes in the intention outcome measures for the different statements, paired samples t-tests were conducted. To determine whether change in drinking intentions significantly differed across messages, effect sizes for each message were calculated based on the within-subject Cohen’s d determined from these t-tests, and 95% confidence intervals for these effect sizes were generated. To examine differences in responses by risk status, which was calculated from respondents’ reported intake levels, independent samples t-tests were conducted. As per the National Health and Medical Research Council’s (NHMRC) Guidelines [24], high short-term risk was defined as drinking more than four standard drinks on a single occasion and high long-term risk was defined as consuming an average of more than two standard drinks per day. Understanding responses according to risk status is important in the light of possible psychological reactance among heavier drinkers [29], which could potentially result in negative unintended consequences from message exposure.

To examine demographic factors associated with attitudinal scores for each statement, a series of linear regressions was conducted. The outcome variables were believability, convincingness and personal relevance. The independent variables used in analyses were gender, age, tertiary education and SES.

To examine factors influencing pre- to post-exposure changes in respondents’ behavioral intentions, multiple linear regression was used across all statements. The outcome variables were (i) beliefs that they should reduce their alcohol consumption, (ii) expectations that they actually will reduce their alcohol consumption and (iii) intentions to drink five or more alcoholic beverages on a single occasion. The independent variables used in analyses were gender (male versus female), age, risk status, socioeconomic status (as per the Australian Bureau of Statistics’ Socio-Economic Indexes for Areas [30] based on residential postcode), education level (tertiary versus non-tertiary), overall believability of the messages and overall personal relevance of the messages. Given the high correlation between believability and convincingness (r = 0.80) and the risk of multicollinearity, only believability was selected for inclusion as a predictor variable in the regression analyses. Regression analyses were accomplished in two steps. First, separate univariate regression analyses were conducted for each possible predictor to avoid any complications due to multicollinearity [31]. In the second step, significant univariate predictors were included in a simultaneous multivariate regression model to determine the unique contribution of each significant predictor.

Results

Pearson chi-square analyses showed that random assignment of participants to warning statements was successful. No significant differences were found between participants randomly assigned to view each message. As such, analyses could proceed as stated above.

Attitudinal outcomes

Table II presents means and standard deviations across all statements and by each statement individually for the outcome variables of believability, convincingness and personal relevance. Statements are ranked by performance. The statement Alcohol increases your risk of bowel cancer was rated highest for believability, convincingness, personal relevance and the attitudinal composite. All six statements received an attitudinal composite score of around 3.5 out of 5, indicating that attitudes to the messages were neutral to favorable. Table III shows the demographic attributes that were significantly associated with responses to individual statements.

Table II.

Attitudinal outcomes by statement

Statement	Believability Mean (SD)	Convincingness Mean (SD)	Personal relevance Mean (SD)	Attitudinal composite Mean (SD)
All statements (N = 1,680)	3.80 (1.00)	3.58 (1.06)	3.06 (1.19)	3.48 (0.90)
Alcohol increases your risk of bowel cancer (n = 285)	3.98 (0.90)	3.74 (0.97)	3.15 (1.17)	3.62 (0.82)
Alcohol increases your risk of breast, bowel, throat and mouth cancer (n = 307)	3.81 (1.05)	3.62 (1.10)	3.13 (1.16)	3.52 (0.94)
Alcohol increases your risk of breast cancer (n = 284)	3.82 (0.93)	3.60 (1.02)	2.98 (1.27)	3.46 (0.83)
Warning: Alcohol increases your risk of cancer (n = 262)	3.71 (1.01)	3.51 (1.11)	3.07 (1.14)	3.43 (0.93)
Alcohol increases your risk of cancer (n = 275)	3.76 (1.06)	3.52 (1.11)	3.00 (1.19)	3.42 (0.94)
Reduce your drinking to reduce your risk of cancer (n = 267)	3.72 (1.03)	3.48 (1.07)	3.01 (1.20)	3.41 (0.90)

Table III.

Demographic factors associated with attitudinal scores by message

Message	Believability	Convincingness	Personal relevance
Alcohol increases your risk of cancer	Age (−)	N/A	N/A
Alcohol increases your risk of breast cancer	N/A	Tertiary education (−)	Gender (females +) Age (−)
Alcohol increases your risk of bowel cancer	N/A	N/A	Gender (males +)
Alcohol increases your risk of breast, bowel, throat and mouth cancer	Gender (females +)	Gender (females +)	Tertiary education (+)
Reduce your drinking to reduce your risk of cancer	N/A	N/A	N/A
Warning: Alcohol increases your risk of cancer	N/A	N/A	N/A

Note: (−) denotes a negative relationship; (+) denotes a positive relationship.

Intention outcomes

Table IV presents post-exposure changes in drinking intentions examined for all statements combined and by each statement individually. Across all the intention outcome variables, favorable differences were found (P < 0.001–0.005; d = 0.44–0.17). Respondents were significantly more likely after exposure to report a perceived need to reduce alcohol consumption and an intention to do so. Overall, the largest effect was evident for the statement Alcohol increases your risk of bowel cancer. All confidence intervals for Cohen’s d overlapped, indicating that the messages were statistically equally effective in reducing drinking intentions.

Table IV.

Alcohol intentions outcomes by statement

Statement(s)	Extent to which should reduce amount of alcohol consumed Mean (SD)			Extent to which actually will reduce amount of alcohol consumed Mean (SD)			Intention to drink 5+ drinks in a single sitting Mean (SD)			Intentions composite Mean (SD)
	Pre- exposure	Post- exposure	Δ	Pre- exposure	Post- exposure	Δ	Pre- exposure	Post- exposure	Δ	Pre- exposure	Post- exposure	Δ	d [95% CI]
All statements (N = 1680)	2.66 (1.29)	2.96 (1.32)***	0.30	2.48 (1.17)	2.76 (1.21)***	0.28	2.64 (1.48)	2.41 (1.37)***	0.23	2.83 (0.78)	3.10 (0.84)***	0.27	0.45 [0.27, 0.52]
Alcohol increases your risk of bowel cancer (n = 285)	2.64 (1.31)	3.04 (1.33)***	0.40	2.52 (1.19)	2.81 (1.22)***	0.29	2.68 (1.50)	2.39 (1.36)***	0.29	2.83 (0.79)	3.15 (0.83)***	0.32	0.54 [0.17, 0.71]
Warning: Alcohol increases your risk of cancer (n = 262)	2.72 (1.22)	2.96 (1.28)***	0.24	2.51 (1.15)	2.82 (1.20)***	0.31	2.66 (1.49)	2.37 (1.36)***	0.29	2.86 (0.76)	3.14 (0.82)***	0.28	0.47 [0.17, 0.64]
Alcohol increases your risk of cancer (n = 275)	2.62 (1.33)	2.88 (1.34)***	0.26	2.43 (1.18)	2.75 (1.27)***	0.32	2.65 (1.45)	2.38 (1.28)***	0.27	2.80 (0.80)	3.08 (0.87)***	0.28	0.45 [0.17, 0.62]
Alcohol increases your risk of breast, bowel, throat and mouth cancer (n = 307)	2.71 (1.32)	3.00 (1.32)***	0.29	2.50 (1.13)	2.77 (1.18)***	0.27	2.68 (1.47)	2.46 (1.39)***	0.22	2.84 (0.77)	3.10 (0.85)***	0.26	0.44 [0.18, 0.60]
Alcohol increases your risk of breast cancer (n = 284)	2.59 (1.32)	2.89 (1.38)***	0.30	2.37 (1.18)	2.63 (1.22)***	0.26	2.57 (1.50)	2.43 (1.47)**	0.14	2.79 (0.77)	3.03 (0.81)***	0.24	0.42 [0.17, 0.59]
Reduce your drinking to reduce your risk of cancer (n = 267)	2.67 (1.27)	2.96 (1.28)***	0.29	2.53 (1.19)	2.78 (1.20)***	0.25	2.60 (1.49)	2.43 (1.38)***	0.17	2.86 (0.81)	3.11 (0.83)***	0.25	0.37 [0.17, 0.54]

^**P < 0.01.

^***P < 0.001.

Predictors of intention outcomes

Of particular interest was whether statement effectiveness varied among different types of drinkers. Additional analyses were conducted to identify respondent attributes associated with changes in the intention outcome variables. Table V provides the significant results from the multivariate regression analyses that were conducted after initial univariate analyses identified respondent attributes that were significantly associated with the outcome variables without the influence of multicollinearity. Given there were no control variables, nor was any predictor of particular interest, all variables were included in one block.

Table V.

Multiple Linear Regression Predicting Intentions

Outcome	Predictor	b	SE	95% CI for b	β	P	Part r2
Believe should reduce (ΔR2 = 0.05)	Believability	0.12	0.02	0.08, 0.17	0.14	<0.001	0.13
	Personal Relevance	0.09	0.02	0.05, 0.13	0.12	<0.001	0.11
Expect will reduce (ΔR2 = 0.05)	Believability	0.07	0.02	0.02, 0.11	0.08	0.003	0.07
	Personal Relevance	0.14	0.02	0.10, 0.17	0.18	<0.001	0.17
Intent to consume 5+ drinks (ΔR2 = 0.02)	Risk profile	−0.09	0.02	−0.14, −0.04	−0.10	<0.001	−0.09
	Believability	−0.05	0.03	−0.10, 0.01	−0.06	0.030	−0.05
	Personal Relevance	−0.04	0.02	−0.09, 0.00	−0.05	0.047	−0.05
Intentions composite (ΔR2 = 0.07)	Believability	0.08	0.02	0.05, 0.11	0.13	<0.001	0.12
	Personal Relevance	0.10	0.01	0.07, 0.12	0.19	<0.001	0.17

Respondents who found the statements more believable and personally relevant reported greater pre- to post-exposure change on each of the dependent variables under examination. Changes in respondents’ intentions to drink five or more drinks in a single sitting were additionally predicted by risk status, with those at higher risk of alcohol-related harm reporting a greater reduction in their intention to drink five or more drinks in a single sitting.

A separate regression analysis was conducted to examine whether the attitudinal composite significantly influenced changes in the intentions composite. This was found to be the case, although the attitudinal composite accounted for only 6.8% of the variance in change in the intentions composite (F(1,1678) = 123.08, P < 0.001, b = 0.18, SE = 0.02, β = 0.26, 95% CI [0.15, 0.21]).

Risk status

Changes in drinking intentions after exposure to the statements were analyzed among those who were at high long-term risk of harm (n = 309, 18% of sample) to identify the statements that were most effective with these high-risk drinkers. When examining the extent to which they believed they should reduce the amount they drink, significant differences pre- to post-exposure were evident for the statements Alcohol increases your risk of bowel cancer (t(52) = 2.97, P = 0.005, d = 0.43) and Reduce your drinking to reduce your risk of cancer (t(47) = 3.09, P = 0.003, d = 0.45). For the extent to which the high-risk drinkers expected that they will reduce the amount they drink, a significant difference pre- to post-exposure was evident for the statements Alcohol increases your risk of breast, bowel, throat and mouth cancer (t(52) = 2.99, P = 0.004, d = 0.41), Alcohol increases your risk of bowel cancer (t(52) = 2.82, P = 0.007, d = 0.39) and Warning: Alcohol increases your risk of cancer (t(47) = 2.63, P = 0.012, d = 0.38).

When examining intention to drink five or more drinks in a single sitting, a significant difference pre- to post-exposure in high-risk drinkers was evident for all statements with the exception of Reduce your drinking to reduce your risk of cancer. The greatest reduction in intention was found for the statements Warning: Alcohol increases your risk of cancer (t(47) = 4.35, P < 0.001, d = 0.63), Alcohol increases your risk of cancer (t(48) = 4.18, P < 0.001, d = 0.60) and Alcohol increases your risk of breast, bowel, throat and mouth cancer (t(52) = 3.29, P = 0.001, d = 0.45).

Finally, a significant change in the intentions composite in high-risk drinkers was evident for all statements. The greatest effect was noted for the statements Alcohol increases your risk of bowel cancer (t(52) = 4.21, P < 0.001, d = 0.59), Warning: Alcohol increases your risk of cancer (t(47) = 4.01, P < 0.001, d = 0.58) and Alcohol increases your risk of breast, bowel, throat and mouth cancer (t(52) = 3.81, P < 0.001, d = 0.52).

Between-group analyses conducted to assess whether drinkers at high risk of long-term harm responded to the statements differently to other drinkers revealed a significant difference in respondents’ assessments of whether they should reduce their consumption after exposure for the message Alcohol increases your risk of breast, bowel, throat and mouth cancer (t(101.963) = 2.03, P = 0.045, d = 0.27). Significant differences were not found for respondents’ assessments of whether they will reduce their consumption after exposure. In terms of intentions, for all statements combined and the statements Alcohol increases your risk of breast, bowel, throat and mouth cancer and Warning: Alcohol increases your risk of cancer, high-risk drinkers reported a significantly greater reduction in intention to consume five or more alcoholic beverages in a single sitting post-exposure relative to other respondents.

Discussion

The results of the present study indicate that cancer warning statements have the potential to encourage drinkers to reduce their alcohol consumption. All six statements assessed in the present study produced favorable outcomes in terms of attitudes to the messages and changes in drinking intentions. This is a promising result given evidence in the tobacco control literature that audience perceptions of message credibility and personal relevance reflect the effectiveness of messages in terms of behavioral change [19, 20].

The identified differences in results between the statements were not large, which is likely to be a reflection of the multi-stage research methodology that involved identification of the most effective messages during an initial testing process [23]. The similar performance of the breast cancer statement, which could be expected to be considered personally relevant by only around half the sample, may be at least partially attributable to previous breast cancer media campaigns that have emphasized that everyone has a mother, sister, or another important woman in their lives and hence that breast cancer touches everyone. The consistent results across the six statements included in the final phase suggest that all these statements would be suitable as part of a suite of rotating warning messages on alcoholic beverages. This approach has been recommended in the light of the successful introduction of rotating warnings on cigarette packets [7]. Where a single statement is required, it appears that the statement Alcohol increases your risk of bowel cancer may be most effective. This may be partially attributable to greater levels of fear arousal resulting from lower perceived survival rates for bowel cancer relative to other common cancers [32, 33].

Regression analyses indicated that, overall, statement effectiveness was not significantly influenced by respondent characteristics. Across the predictor variables assessed in this study, believability and personal relevance were the largest contributors to statement effectiveness. These outcomes highlight the importance of public education about the association between alcohol consumption and cancer and the promotion of the current national recommendations relating to low-risk drinking [24]. Such education is needed to maximize drinkers’ receptiveness to the statements and enhance awareness of the personal relevance of the messages being conveyed.

Of note is that high-risk drinkers had very similar responses to the warning statements as other drinkers, suggesting that the same messages can be effective among the general population of drinkers and those who are most at risk of alcohol-related harm. This outcome may serve to allay concerns relating to possible psychological reactance among heavy drinkers [29], which could potentially result in the contra-indicated behavior becoming more attractive [34]. The results of the present study suggest that heavy drinkers will be just as, if not more, likely to exhibit the desired response to the messages.

The modest ability of attitudes to predict intentions suggests that future research should prioritize intention measures over attitudinal measures when assessing warning statement effectiveness. The need to move beyond collecting attitudinal data was also apparent in the differences in statement ranking according to the attitude composite score compared with the intentions composite score. This outcome supports previous work demonstrating that the most preferred health messages are not necessarily the most effective [35].

A limitation of the present study was the use of a web panel provider to recruit the sample of drinkers, which means population representativeness cannot be assumed. In addition, the response rate cannot be calculated because respondents could either reply to an invitation email or spontaneously access the web panel portal to search for surveys for which they were eligible to participate. However, the primary aim of the study was to test the relative effects of different message options rather than estimating population parameters. Future research could seek to use population representative samples, which would ideally also include non-drinkers and underage drinkers to assess message effects in terms of reinforcing non-consumption and delaying or dampening alcohol initiation. To assess whether the study design influenced outcomes, future work in this area could also include a control group that is exposed to a non-alcohol-related message. Finally, further research is needed to determine whether these results found among a large sample of Australian drinkers are replicated in other cultural contexts.

Conclusion

It is recognized that warning statements in isolation cannot be expected to result in substantial changes in behavior and instead require reinforcement via other forms of information provision, such as via general practitioners, the mass media and inclusion in the education curriculum [24, 36]. The results of the present study demonstrate that cancer warning statements have the potential to play an important role in public education programs designed to inform drinkers of the long-terms harms associated with alcohol consumption and encourage behavioral change. Such programs are likely to be more effective if delivered across a range of contexts. The mass media, product packaging, medical practitioner and word-of-mouth information delivery mechanisms represented in the simulation are all likely to be important elements of any comprehensive alcohol control strategy. The warning statements tested in this study appear suitable for delivery across these varying information dissemination contexts.

Funding

The work was supported by the Western Australian Health Promotion Foundation (Healthway), research grant 20338.