Type of message and framing

Thomson et al. (2012) compared consumer responses to serious and humorous alcohol warning labels of various formats during focus groups.  There findings suggest that humorous messages are not effective as they were less accepted by participants because of a lack of comprehension of the message.
Glock and Krolak-Schwerdt (2013) explored university students' reactions to five health-related (health) warnings and five (social) warnings countering expectations of positive social and relaxing utilities of alcohol. With one group of participants exposed to the health and one group exposed to the social warnings. Findings showed that explicit attitude towards drinking alcohol did not differ across the two groups. However the participants differed in their implicit attitude with those in the social warning label group having less positive attitudes towards alcohol and those in the health group having more positive implicit attitudes.

Jarvis and Pettigrew (2013) explored negative and positive framed messages (across both health and drink driving contexts) and found that for those who report higher consumption of alcohol, negative health messages had the highest utility. A positive statement about drink driving exhibiting a boomerang effect and positively impacting choice suggesting higher likelihood of purchasing alcohol.

Collymore and McDermott (In press) examined social and health messages across gain, fear-loss and disgust-loss conditions, and found that loss-framed messages (and in particular the health disgust-loss with a message about drinking more than government daily limits results in temporary facial imperfections) were most effective in increasing intentions to reduce alcohol consumption.

 

Conclusion: It is difficult at this point to advocate a particular type of message.  This is because there exists little consistent findings with a lack of replication studies limiting the confidence to be gained from the body of knowledge formed to date.  Studies so far are based on exploratory methods (such as focus groups), or descriptive studies that utilised small samples, or within designs (by repeated presentations of different warnings) whilst failing to rule out order (carry-over) effects. As a result the field needs more studies that will in time allow a meta-analysis to be conducted to enable a more rigorous assessment of the utility of different types of messages. In the meantime researchers should continue to explore both positively and negatively framed messages and examine potential mediators (e.g., perceived susceptibility) and moderators (e.g., believability) of behavioural change as well as assess for potential defensive reactions or other boomerang effects.

 

Slater et al. (1998) examined the influence of providing a recommendation (in addition to a warning) and found no effect on the number of negative responses or risk perceptions across three messages – drinking and driving, health, alcohol & drugs. There was an effect on believability and number of positive responses such that the health and alcohol & drugs messages were more believable with a recommendation, yet an opposite effect was found for the drink driving message. Pettigrew et al. (2014) evaluated general and specific cancer-related alcohol warning messages and found that two positively worded messages both phrased as a recommendation to reduce drinking alcohol were rated by respondents as more believable than negatively worded messages that relied on fear arousal and didn’t provide a recommendation. 

 

Conclusion: There is a need for more research regarding the influence of providing a recommendation across different types of messages.  These results suggest that providing a recommendation may be beneficial but the efficacy of a recommendation needs to be tested experimentally across a range of different message themes to make a firmer decision on the utility of recommendations.

 

Slater et al. (1998) found that recall of the message was improved with the use of quantitative (as opposed to qualitative) information in the warning label.  However quantitative information was less likely to be believed for two of the three messages tested (drinking and driving and health). Further Pettigrew et al. (2014) included two negative quantitative messages to evaluate against 10 other messages (positive, negative fear arousal, general or specific cancer message) and found that the quantitative messages performed poorly in terms of believability and was influenced by gender.

 

Conclusion: It appears from the limited evidence available that quantitative messages are of less utility than qualitative messages but researchers should evaluate the efficacy of quantitative information across a range of message types and frames. In addition to exploring subgroup effects across psychological variables such as need-for-cognition, and demographic factors.