Building a Better Message: The 10 Variables That Really Matter (The Research)
Punam Keller, PhD, MBA
Tuck School of Business at Dartmouth College, Hanover, NH
Dr. Keller explores extensive meta-analysis of the main and interaction effects of message tactics and individual
characteristics on intentions to comply with health recommendations. Based on her research, Dr. Keller discusses
the empirical model on which the Message Development Tool is based and the 10 variables that are significant
predictors for stated intentions and behavior when socioeconomic, social influence, beliefs and attitudes, number
of ads, and exposure frequency are accounted for.
5. METHODS AND RESULTS BACKGROUND CONCLUSIONS IMPLICATIONS FOR PRACTICE Division of Cancer Prevention and Control Table 2: Effective Matches between Message Tactics and Audience Characteristics from the Keller and Lehmann Advisor for Risk Communication Model All ages respond to messages advocating detection behaviors Nonwhites seem to care more about vivid messages that emphasize the effect of health consequences on loved ones Women respond to emotional messages with social consequences for themselves or health consequences to near and dear ones Men are more influenced by unemotional messages that emphasize personal physical health consequences Contrary to popular use, framed health messages (loss or gain frames) are not advisable without knowledge of target audience goals (promotion vs. prevention)
13. Ad Exposures Commercial Pattern 1 Pattern 2 Pattern 3 Pattern 4 Bike Race 1 1 1 1 Dribbling 0 1 1 1 Life Guard 0 0 1 1 Venus Williams 0 0 0 1 Exponent of Sum Rule .77 .81 .86 .93 Intention Max Rule .77 .77 .77 .78 Sample Size 184 309 572 104
14.
15.
16.
17.
Hinweis der Redaktion
Theoretical Background and research questions/hypothesis: Currently, four barriers prevent the application of research to improve the effectiveness of public health communication campaigns. First, the focus on one or two message tactics makes it difficult to generalize the results to situations where the audience is faced with a wide variety of message tactics in the same or different health campaigns. Second, most health communication studies do not provide guidelines for tailoring since they do not examine how message formats interact with measurable individual differences such as psychographics. Third, small sample sizes in most studies raise concerns about whether findings can be replicated in the field. Finally, there is no evidence that message formats determine health intentions when other factors such as peer influence are taken into account.
Methods and Results (informing the conceptual analysis): To address this need, Drs. Keller and Lehmann systematically examined the role of message tactics and individual differences on intentions to comply with health recommendations. A meta-analysis of 60 experimental studies, involving 584 health message conditions and 22,500 participants, was conducted to examine main and interaction effects of 22 message tactics (e.g. gain/loss framing, vividness, self/other referencing, emotion) and six individual characteristics (e.g. gender, age, race, involvement) on intentions to comply with health recommendations (Keller and Lehmann, 2008). The authors used two approaches to identify matches between message tactics and audience characteristics: a full and a reduced regression model.
Methods and Results (informing the conceptual analysis)(continued):
Methods and Results (informing the conceptual analysis)(continued):
Methods and Results (informing the conceptual analysis)(continued): Keller and Lehmann's research suggests an empirical model to tailor health communications for different target audiences to increase intentions. Results were further validated through application to the CDC Verb campaign (2004-2006), a process which involved 1.) coding CDC Verb campaign advertisements; 2.) using the model to calculate intention and behavior estimates; and 3.) comparing the model estimates to extensive evaluation data collected on outcomes of the Verb campaign. Keller and Lehman tested the model with CDCâs VERB campaign and found that the ARC predictions and stated intentions are closely correlated when socioeconomic status, social influence, beliefs and attitudes, number of ads, and exposure frequency are accounted for.
They also asked the parents a variety of socio-economic and demographic questions. We can use ARC to calculate the intentions to exercise in response to each ad â for example the same biike racing ad varies in effectiveness with lower effectiveness for whites than non-whites.
The model is based on a single exposure to a single ad. However, in reality, the kids reported seeing multiple VERB ads. Although there are several other possibilities, I will show you the results of two models, one where we add the intentions and smooth the curve using an exponential model. The second, where we select the most effective ad seen according to the ARC model. We will call them the exponent of the sum and the max rule respectively.
Conclusions: Keller and Lehmann's research suggests an empirical model to tailor health communications for different target audiences. Keller and Lehmann's empirical model provides 10 variables that are significant predictors for stated intentions and behavior when socio-economic, social influence, beliefs and attitudes, number of ads, and exposure frequency are accounted for. Intention and behavior predictions are approximately equally sensitive to family and social influence, parent education, and recall of message exposures, and in general have less impact than the child variables or model predictions.
Implications for research and/or practice: Â Results show there is a significant opportunity to tailor health communications and even market public health more efficiently to different market segments. Keller and Lehmann's (2008) model formed the basis for CDC DCPC's Message Development Tool (MDT).