Research data supporting "Impact of increasing vegetarian availability on meal selection and sales in cafeterias"

Abstract

This research data contains data from across 2017 on vegetarian sales, total sales, summaries of individual meal selections and model estimates from cafeterias at University of Cambridge colleges A, B and C. The aim of the study was to test investigate the impact of increasing vegetarian availability (as a proportion of total options) on vegetarian sales and total sales. We collected data from 3 University of Cambridge college cafeterias during weekday term-time lunches and dinners (the university’s colleges are broadly equivalent to halls of residence). All colleges already varied the number of total meal options and vegetarian (including vegan) options served at lunch and dinner. Study 1 comprised nonexperimental data of 86,932 hot main meals (salads and sandwiches were not included) from colleges A and B across lunch and dinner during the spring, summer, and autumn terms in the 2017 calendar year. Study 2 consisted of experimental data of 7,712 meals from college C lunches during the autumn term in 2017, when we experimentally altered the number of vegetarian options on offer at lunchtimes, between 1 and 2. We summarized the sales transaction data into 1) aggregate data, summarizing the total vegetarian and meat/fish (hereafter “meat”) sales at each lunch and dinner and 2) individual-level data on whether each diner at a meal selected a vegetarian or meat meal. Purchases made with university cards enabled anonymized individual diner-level purchases to be tracked; this is useful in evaluating how diners with different pre-study levels of purchasing vegetarian meals responded to increasing vegetarian availability. We used the total number of vegetarian and meat meals sold at a mealtime to analyse total sales. Measuring rebound effects (i.e., increased meat purchases at another time) is not possible for study 1 as vegetarian availability varied across lunches and dinners. For study 2, although we cannot completely capture rebound effects as we do not have information on what diners ate outside the cafeteria, as a proxy, we measured vegetarian sales at college C during dinnertimes, which were not included in the experimental intervention. We had originally intended dinners to be included, but this posed too much of an operational burden for the cafeteria. This created the opportunity to conduct a post hoc analysis of rebound effects that was not part of the original study design. We estimated the effect of vegetarian availability on vegetarian meal sales and total meal sales, adjusting for other predetermined variables, including day of the week, ambient temperature, and average price difference between vegetarian and meat options using linear models (LMs) and binomial generalized linear models (GLMs) for aggregate data. Binomial generalized linear mixed models (GLMMs) were used for the individual-level data, with individual diner fitted as a random effect, which allows each diner to have a different likelihood of selecting a vegetarian meal. A 95% confidence level was used to calculate confidence intervals (CIs). Models were evaluated using the Akaike information criterion (AIC), interpretability, and model diagnostics. $$ $$ A minor amendment was made to the dataset originally deposited in the repository (Garnett_2019_PNAS_ResearchData_VegetarianAvailability.xlsx) - this dataset version is no longer available for download. The amended dataset (Garnett_2019_PNAS_ResearchData_VegetarianAvailability-2.xlsx) was uploaded on 6 January 2020.