The Effect of Graphic Warnings on Sugary-Drink Purchasing

Abstract

Governments have proposed text warning labels to decrease consumption of sugary drinks-a contributor to chronic diseases such as diabetes. However, they may be less effective than more evocative, graphic warning labels. We field-tested the effectiveness of graphic warning labels (vs. text warning labels, calorie labels, and no labels), provided insight into psychological mechanisms driving effectiveness, and assessed consumer sentiment. Study 1 indicated that graphic warning labels reduced the share of sugary drinks purchased in a cafeteria from 21.4% at baseline to 18.2%-an effect driven by substitution of water for sugary drinks. Study 2 showed that graphic warning labels heighten negative affect and prompt consideration of health consequences. Study 3 indicated that public support for graphic warning labels can be increased by conveying effectiveness information. These findings could spur more effective labeling policies that facilitate healthier choices, do not decrease overall beverage purchases, and are publicly accepted.

Trial registration: ClinicalTrials.gov NCT02744859 NCT02947802.

Keywords: decision making; food; health; open data; open materials; policy making; preferences; preregistered.

Conflict of interest statement

Declaration of Conflicting Interests: The author(s) declared that there were no conflicts of interest with respect to the authorship or the publication of this article.

Figures

Fig. 1.

Labels used in Studies 1, 2, and 3: (a) calorie label, (b) text warning label, and (c) graphic warning label. Studies 1 and 3 used all three labels, whereas Study 2 used only the graphic warning label.

Fig. 2.

Sequential mediation of the effect of graphic warning labels on the increased intention to purchase water instead of a sugary drink via negative affect and health consideration (Study 2). Path coefficients are standardized regression weights. On the path from graphic warning label to intention to purchase water, the coefficient above the line represents the direct effect without the mediators in the model, and the coefficient below the line represents the direct effect with the mediators in the model. Asterisks indicate significant paths (**p < .01, ***p < .001).

Fig. 3.

Mean rating of consumer support of labels by effectiveness information and label type in Study 3 (N = 402). Error bars indicate ±1 SEM.