The steaks are high: Reducing meat consumption by changing physical and economic environments to increase vegetarian sales

Livestock farming is responsible for ~15% of anthropogenic greenhouse gas emissions (GHGE) and is a leading cause of deforestation, biodiversity loss, water-use and pollution. To feed 10 billion people a healthy and sustainable diet, approximately 16kg of meat consumption per person per year has been recommended, compared with current mean consumption (including waste) of 81kg in the EU and UK. How might lower meat consumption be encouraged? Research in other domains suggests altering the physical and economic environments in which people make decisions holds promise for achieving socially desirable behaviour change, but very little research has experimentally tested such approaches for reducing meat consumption. In this thesis I present the results of three different interventions (order, availability, price) in college cafeterias at the University of Cambridge and examine their effects on vegetarian meal sales. I collected data from 1142 mealtimes and 213,627 meal selections, obtaining individual level purchase information for two out of three interventions. It is widely assumed – but largely untested – that food encountered first in cafeterias is preferentially selected. I investigated the effects of order and found placing the vegetarian (rather than meat) option first increased vegetarian sales by 4.5 to 6 percentage points when there was a long distance (181cm) between options. However, order effects were inconsistent when the distance between options was shorter (<85cm). In contrast, I found that increasing the proportion of vegetarian options available was consistently very effective. Doubling vegetarian availability from 25% to 50% (e.g. from 1 in 4 to 2 in 4 options) increased vegetarian meal sales by 14.5 to 14.9 percentage points in an observational study and by 7.8 percentage points in an experimental study. Individual-level data revealed that the largest relative effects were found in the quartile of diners with the lowest prior levels of vegetarian meal selection, but all quartiles of diners were more likely to select a vegetarian option when more were available. Price is an important consideration for citizens when purchasing food. I experimentally decreased the vegetarian option price and increased the meat option price (each by 20p) halfway through a university term. Vegetarian sales increased overall by 3.2 percentage points, and by 13.7 percentage points in the most vegetarian quartile of diners. The other three quartiles did not significantly change their meal selections. None of the three interventions tested substantially affected overall meal sales. In the final data chapter I used individual-level data to examine the effects of gender on meal selection. I found that men were consistently less likely to select vegetarian meals than women, significantly more likely to select meat meals, and men and women were equally likely to select fish meals. Consequently on average men’s meals had average GHG emissions 18% higher and land-use 28% higher than women’s. Men and women were similarly responsive to the availability and price interventions. These findings have important implications for catering policies, although these interventions should be tested in non-university populations and low and middle income countries. Placing vegetarian options first can increase their sales, but can also have no effect or even be counterproductive. A small change in price may only be enough to increase vegetarian selection for the most vegetarian quartile of diners. However, increasing the availability of vegetarian options appears to increase vegetarian selection by all quartiles of diners and is a relatively simple change to catering practices. My results provide robust evidence that – if implemented more broadly – increasing the proportion of vegetarian options available could make an important contribution to the global ambition for more sustainable diets.