The Role of Narrative Pictorial Warning Labels in Communicating Alcohol-Related Cancer Risks

ABSTRACT

This study examined the role of pictorial warning labels (PWLs) featuring narrative content in communicating alcohol-related cancer risks. In an online experiment, 169 adult alcohol consumers were randomly assigned to view two narrative PWLs, two non-narrative PWLs, or control. Results showed that exposure to narrative PWLs significantly increased participants’ worry about, feelings of risk of, and perceived severity of harm of getting alcohol-related cancer, but did not affect their comparative likelihood of getting alcohol-related cancer or intentions to reduce alcohol use. Exposure to narrative PWLs also indirectly influenced intentions through increased worry. Moreover, participants’ risk perceptions and intentions in non-narrative PWLs condition did not differ from those in narrative PWLs and control conditions. Therefore, these findings suggest that narrative PWLs are a promising strategy in informing consumers about the cancer risks of alcohol.

Notes

1. Previous research showed that PWLs had a medium effect on outcomes such as alcohol-related cancer risk perceptions (d = 0.53 in Clarke et al., 2021; η² = .19 in Wigg & Stafford, 2016). With alpha level set at .05 and effect size f set at 0.25, a sample size of 159 will achieve a statistical power of .80.

2. Although feelings of risk were highly correlated with worry (r = .85) and with comparative likelihood (r = .82), the variance inflation factor (VIF) value for each type of risk perception was between 1 and 5, suggesting that multicollinearity was not an issue in the regression model.

3. This study also measured anger, the affective component of reactance. Although the descriptive statistics showed that anger in the narrative PWLs condition (M = 3.68, SD = 1.69) was slightly lower than it in the non-narrative PWLs condition (M = 3.84, SD = 1.63), the difference was not statistically significant, t(111) = .53, p = .60. A possible reason is that the study sample size was relatively small, and thus, it lacked the power to detect significant differences. As previous research suggested that narrative had a small effect on reactance (Ratcliff & Sun, 2020), a sample size of 788 will be needed to achieve a statistical power of .80 when the alpha level is set at .05 and effect size at .20.