Skip to main content
. 2019 Sep 30;116(42):20923–20929. doi: 10.1073/pnas.1907207116

Fig. 3.

Fig. 3.

Effects of vegetarian availability on vegetarian and total sales for college C, study 2. (A) Raw values of vegetarian sales against vegetarian availability. (B) Modeled likelihood of selecting a vegetarian meal for individual diners, divided into least vegetarian (LeastVeg) to most vegetarian (MostVeg) quartiles. LessVeg, less vegetarian; MoreVeg, more vegetarian. (C) Raw values of total sales against vegetarian availability. (D) Raw values of vegetarian sales at dinner during the control and experimental weeks, with model mean estimates and CIs shown in white. Lines of best fit and CIs in A and C and model mean estimate with CIs in D were generated from the models using conditional regression and the visreg package in R (Methods).