PROJECT SUMMARY Poor dietary habits are associated with obesity, type 2 diabetes, cardiovascular disease, and certain cancers. Government and institutional policy makers worldwide are interested in food labeling and messaging as a cost-effective strategy to encourage healthier food choices. Although there is evidence that nutrition labels can influence behavior, it is unclear which label designs are most effective at altering behavior for people with different levels of education. There are three common types of nutrition labels: a single logo highlighting healthier choices, traffic light labels, or physical activity labels that translate calorie information into exercise required to burn off the calories. Because there have been no large-scale, long-term randomized trials comparing the effect of these labels on behavior, the relative impact of these labeling schemes is unknown. Such rigorous evaluations are needed so policy makers do not have to guess. The overall objective of this application is to compare four theoretically-informed nutrition label designs in the first long-term, large-scale randomized trial of nutrition messages. The study will be conducted in partnership with the Philadelphia Department of Public Health which will enable us to randomize 267 (107 snack and 160 beverage) vending machines to different nutrition label messages. Beverage machines will be randomized to one of the following four conditions: 1) A single green traffic light label next to items meeting a “healthy” threshold; 2) a multiple traffic light label that displays red, yellow, or green labels to indicate options that range from less to more healthy; 3) a physical activity label that indicates the number of minutes one would have to walk to burn off the calories in each food or beverage; or 4) tax-related messaging (Philadelphia implemented a 1.5 cents per ounce beverage tax in January 2017). Snack machines will be randomized to the same conditions except for the tax messaging, which does not apply to foods. We will use vending machine transaction data over three years to compare the influence each label has on average calories and percentage of healthy and less healthy items purchased as well as impact on sales. We will also conduct interviews with 2,916 vending machine patrons to evaluate whether education level and other factors moderate message influence. This study will provide valuable data to fill research gaps and inform policy makers’ efforts to use nutrition labeling and messaging as a public health approach to encourage healthier choices and address health disparities. !