The Effect of Calorie Posting Regulation on Consumer Opinion: A Flexible Latent Dirichlet Allocation Model with Informative Priors

In 2008, New York City mandated that all multi-unit restaurants post calorie information in their menu. For managers of multi-unit and stand-alone restaurants, and for policy makers, a pertinent goal might be to monitor the impact of this regulation on consumer conversations. We propose a scalable Bayesian topic model to measure and understand changes in consumer opinion about health (and other topics). We calibrate the model on 761,962 online reviews of restaurants posted over 8 years. Our methodological contribution is to generalize topic extraction approaches in marketing and computer science. Specifically, our model does the following: a) each word can probabilistically belong to multiple topics (e.g. “fries” could belong to the topics “taste” as well as “health”); b) managers can specify prior topics of interest such as “health” for a calorie posting regulation; and c) review lengths can affect distributions of topic proportions so that longer reviews might include more topics. Through careful controls, we isolate the potentially causal effect of regulation on consumer opinion. Following the regulation, there was a small but significant increase in the discussion of health topics. Health discussion remains restricted to a small segment of consumers. * Dinesh Puranam is a Marketing Ph.D. student at Johnson, Cornell University. This paper is part of his doctoral thesis. Vishal Narayan is an assistant professor of Marketing at NUS, Singapore. Vrinda Kadiyali is the Nicholas H. Noyes Professor of Marketing and Economics at Johnson, Cornell University. They can be reached at [email protected], [email protected] and [email protected] respectively.