Assessing Consumer Buy and Pay Preferences for Labeled Food Products with Statistical and Machine Learning Methods

Abstract

ABSTRACT Food labeling is one approach to encourage safe, healthy, and sustainable dietary practices. Consumer buy and pay preferences for specially labeled food products (e.g., U.S. Department of Agriculture organic, raised without antibiotics, and locally raised) may promote the adoption of associated production practices by food producers. Thus, it is important to understand how consumer buy and pay preferences for specially labeled products vary with their demographics, food-relevant habits, and foodborne disease perceptions. Using both conventional statistical and novel machine learning models, this study analyzed Michigan State University Environmental Science and Policy Program annual survey data (2019) to characterize consumer buy and pay preferences regarding eight labels related to food production practices. Older consumer age was significantly associated with lower consumer willingness to pay more for labeled products. Participants who prefer to shop in nonconventional grocery stores were more willing to buy and pay more for labeled products. Our machine learning models provide a new approach for analyzing food safety and labeling survey data and produced adequate average prediction accuracy scores for all eight labels. The label “raised without antibiotics” had the highest average prediction accuracy for consumer willingness to buy. Thus, the machine learning models may be used to analyze food survey data and help develop strategies for promoting healthy food production practices. HIGHLIGHTS