ArticlePDF Available

Impact of increasing vegetarian availability on meal selection and sales in cafeterias

Authors:

Abstract and Figures

Shifting people in higher income countries toward more plant-based diets would protect the natural environment and improve population health. Research in other domains suggests altering the physical environments in which people make decisions (“nudging”) holds promise for achieving socially desirable behavior change. Here, we examine the impact of attempting to nudge meal selection by increasing the proportion of vegetarian meals offered in a year-long large-scale series of observational and experimental field studies. Anonymized individual-level data from 94,644 meals purchased in 2017 were collected from 3 cafeterias at an English university. Doubling the proportion of vegetarian meals available from 25 to 50% (e.g., from 1 in 4 to 2 in 4 options) increased vegetarian meal sales (and decreased meat meal sales) by 14.9 and 14.5 percentage points in the observational study (2 cafeterias) and by 7.8 percentage points in the experimental study (1 cafeteria), equivalent to proportional increases in vegetarian meal sales of 61.8%, 78.8%, and 40.8%, respectively. Linking sales data to participants’ previous meal purchases revealed that the largest effects were found in the quartile of diners with the lowest prior levels of vegetarian meal selection. Moreover, serving more vegetarian options had little impact on overall sales and did not lead to detectable rebound effects: Vegetarian sales were not lower at other mealtimes. These results provide robust evidence to support the potential for simple changes to catering practices to make an important contribution to achieving more sustainable diets at the population level.
Content may be subject to copyright.
This is the authors’ accepted version of the manuscript for personal use. The definitive version is published in
PNAS, www.pnas.org/cgi/doi/10.1073/pnas.1907207116 . Supporting information online at
www.pnas.org/lookup/suppl/doi:10.1073/pnas.1907207116/-/DCSupplemental .
1
Classification
SOCIAL SCIENCES: Sustainability Science
Title
Impact of increasing vegetarian availability on meal selection and sales in cafeterias
Authors
Emma E. Garnetta,1
Andrew Balmforda
Chris Sandbrookb
Mark A. Pillingc
Theresa M. Marteauc
a Department of Zoology, University of Cambridge, CB2 3EJ, UK; b Department of Geography,
University of Cambridge, CB2 1QB, UK; c Behaviour and Health Research Unit, University of
Cambridge, Institute of Public Health, CB2 0SR, UK
1 To whom correspondence may be addressed. Email: ee.garnett@gmail.com
Keywords
Diet; behaviour change; meat; vegetarian; choice architecture; sustainability; climate
change; cafeterias
Data deposition
The aggregate data and summaries of the individual-level data can be found at
https://doi.org/10.17863/CAM.41328 .
2
Abstract
Shifting people in higher-income countries towards more plant-based diets would protect
the natural environment and improve population health. Research in other domains
suggests altering the physical environments in which people make decisions (“nudging”)
holds promise for achieving socially desirable behaviour change. Here we examine the
impact of attempting to nudge meal selection by increasing the proportion of vegetarian
meals offered in a year-long large-scale series of observational and experimental field
studies. Anonymised individual-level data from 94,644 meals purchased in 2017 were
collected from three cafeterias at an English university. Doubling the proportion of
vegetarian meals available from 25% to 50% - e.g. from 1 in 4 to 2 in 4 options - increased
vegetarian meal sales (and decreased meat meal sales) by 14.9 and 14.5 percentage points
in the observational study (two cafeterias) and by 7.8 percentage points in the experimental
study (one cafeteria), equivalent to proportional increases in vegetarian meal sales of
61.8%, 78.8% and 40.8% respectively. Linking sales data to participants’ previous meal
purchases revealed that the largest effects were found in the quartile of diners with the
lowest prior levels of vegetarian meal selection. Moreover serving more vegetarian options
had little impact on overall sales and did not lead to detectable rebound effects: vegetarian
sales were not lower at other mealtimes. These results provide novel and robust evidence
to support the potential for simple changes to catering practices to make an important
contribution to achieving more sustainable diets at the population level.
Significance statement
Reducing meat consumption in higher income countries is vital to protect the environment
and improve public health. Few studies have tested the real-world performance of different
strategies to increase plant-rich diets, and none has examined the impact of altering the
availability of vegetarian meal options. In robust observational and experimental studies, we
show that doubling the proportion of vegetarian meals offered increases vegetarian sales by
between 41% and 79%. Our study is the first study to assess the impact of increasing the
proportion of plant-based meal options on selection, and is based on over 90,000 meal
choices. We suggest our findings have potential to make a significant contribution to the
global ambition for more sustainable diets.
3
Introduction
High-income countries produce and consume animal-derived food meat, fish, dairy, eggs
at levels that are incompatible with meeting greenhouse gas emissions (GHGE) reduction
targets (1). Livestock and aquaculture are responsible for 56-58% of the global food system’s
GHGE and use 83% of farmland despite contributing just 18% of calories and 37% of our
protein (2). In particular, meat from ruminants (cows, sheep, goats) has average GHGE per
kg five times higher than pork, seven times higher than chicken and 43 times higher than
legumes (3). Shifting towards a more plant-based diet is therefore one the most effective
ways of reducing the environmental footprint of food (2, 4). For the UK it is estimated that
switching from a high meat (>100g/day) to an entirely vegetarian diet would reduce the
GHGE of a typical person’s food by 47% (5).
Shifting diets to achieve sustainability outcomes is likely to require an array of strategies for
changing human behaviour (6, 7). Education to bring about behaviour change is a popular
and uncontroversial method but while it can raise awareness it appear to be largely
ineffective at actually changing behaviour (8, 9). Models suggest that taxes on the most
polluting foods would result in savings of 1Gt of GHGE worldwide (4) but these taxes can be
regressive and are politically unpopular given their lack of public support (8). A third group
of interventions changing the physical, economic and social context (the so-called choice
architecture) in which decisions are made could potentially deliver improved
environmental outcomes at a low cost and with little controversy, but so far has received
relatively little empirical attention (1013).
As one form of nudging, altering the relative availability of different food types has shown
promise as a lever for changing dietary behaviour to improve population health. Reducing
the availability of high calorie foods is estimated to be the third most effective strategy for
combatting obesity after lowering portion size, and reformulation, although the evidence
for subsequent behaviour change is rated as “limited” (14). A Cochrane review (15) found
only five studies on altering availability that met the inclusion criteria (1620), with a meta-
analysis showing a non-significant decrease in consumption and a large significant decrease
in selection. Other studies on availability, not included in the Cochrane review, have found
increasing the relative availability of low- and moderate-fat entrées in a USA school
4
cafeteria from 33% to 50% increased their selection by 108% and 63% respectively (21); and
in four English workplace cafeterias, decreasing the number of high-calorie cooked meals
offered to one option per lunchtime (while keeping the total number of options offered
constant) reduced the mean energy per main meal sold by 26.1% (22).
Turning to reducing meat consumption, a recent review found no studies on the effects of
changing the availability of plant-based meals (13). The likely patterns are hard to
anticipate: at one extreme increasing relative availability might have a directly proportional
impact on relative sales; conversely, if people have fixed preferences for meat or vegetarian
meals, changing their relative availability might have no impact. It is important in such work
that outcomes are assessed over sustained periods, because effects can wane over time (23,
24), and if possible that inter-individual variation is examined too: an online study altering
menu configurations found different responses between those who frequently or
infrequently ate vegetarian foods (25). However, we are aware of only one study (again
focused on health rather than meat consumption) which presents long-term individual-level
data on how availability affects food choices (26). There are two further considerations: for
any intervention to be acceptable to caterers, it is important that total sales and revenue do
not substantially drop as a result (24, 27); and to have a genuinely additional environmental
effect it is important there are no sizeable rebound effects (28) whereby meat consumption
increases on other occasions. However almost no studies address rebound effects or effects
on total sales (24).
To tackle these research gaps, we conducted two studies one observational and one
experimental in three college cafeterias in the University of Cambridge. These studies
examined the effect on vegetarian sales of increasing the proportion of vegetarian options
available (hereafter “availability”). We tested the hypothesis that meal selection is
influenced by availability, such that increasing the availability of vegetarian options
increases their selection. In these studies we take advantage of year-long and anonymised
individual-level data to analyse whether increasing vegetarian availability had effects which
differed with the prior levels of vegetarian meal consumption of individual diners, affected
total sales, or resulted in rebound effects at other mealtimes when vegetarian availability
was not altered.
5
Research setting
We collected data from three 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
options served at lunch and dinner. Vegetarian options contained no meat or fish, but may
have included eggs and dairy products; vegan options were entirely plant-based, and
therefore contained no eggs or dairy products. Approximately 30% of the vegetarian options
on offer were vegan. Hereafter vegetarian and vegan options are both referred to as
“vegetarian”. Study 1 comprised non-experimental data of 86,932 hot main meals
(hereafter referred to simply as “meals”; salads and sandwiches were not included) from
Colleges A and B, across lunch and dinner during spring, summer and autumn terms in the
2017 calendar year (Figure 1). Study 2 consisted of experimental data of 7712 meals from
College C lunches during autumn term 2017, when we experimentally altered the number of
vegetarian options on offer at lunchtimes (Figure 1).
We summarised the sales transaction data into a) aggregate data, summarising the total
vegetarian and meat/fish (hereafter simply “meat”) sales at each lunch and dinner and b)
individual-level data on whether each diner at a meal selected a vegetarian or meat meal.
Purchases made with university cards enabled anonymised 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 (Methods). 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 dinner times, 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 (Methods). This created the opportunity to conduct a post-hoc
analysis of rebound effects that was not part of the original study design.
6
We estimated the effect of vegetarian availability on vegetarian meal sales and total meal
sales, adjusting for other pre-determined variables including day of the week, ambient
temperature, average price difference between vegetarian and meat options (Methods)
using Linear Models (LMs) and binomial Generalised Linear Models (GLMs) for aggregate
data. Binomial Generalised 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 (29). A 95% confidence level was used to
calculate confidence intervals (CIs). Models were evaluated using the Akaike Information
Criterion (AIC), interpretability and model diagnostics (30).
Figure 1: Overview of data and levels of analyses in Study 1 and Study 2. Credit: icons from
thenounproject.com.
7
Study 1: observational
Aims and design
For Study 1 we did not experimentally alter the menu (Supporting Information (SI)
Appendix, Tables S1 and S2) but observed the number of vegetarian and meat options
available from the sales data. We analysed long-term data from 269 mealtimes at College A
and 266 mealtimes at College B. Excluding the few mealtimes where no vegetarian options
were served (SI Appendix Tables S3 and S4), vegetarian availability ranged from 16.7% to
75% in College A and 12.5% to 66.7% in College B.
Vegetarian sales: aggregate data
Vegetarian availability alone explained 20.9% and 31.9% of variation in vegetarian sales at
College A and College B respectively (Binomial GLMs, McFadden’s pseudo R2). When
controlling for other variables the best GLMs for College A and B explained 26.1% and 39.3%
respectively of the variability in vegetarian sales (SI Appendix Tables S5 and S6), with
vegetarian availability remaining a highly significant predictor of vegetarian sales for both
colleges (College A, n= 51,251 meals, p<0.001; College B, n= 35,681 meals, p<0.001).
Specifically, the models estimated that doubling vegetarian availability from 25% to 50%
increased vegetarian sales by 61.8% in College A (from 24.1% (CI= 22.5%, 25.7%) to 39.0%
(CI= 36.7%, 41.3%) of total sales) and by 78.8% in College B (from 18.4% (CI= 16.8%, 20.1%)
to 32.9% (CI= 30.6%, 35.4%), Figure 2a and SI Appendix Tables S5 and S6).
Other variables also correlated with vegetarian sales but often had different effects in the
two colleges. For example, as the vegetarian option became relatively cheaper compared to
the meat options, vegetarian sales increased in College A but decreased in College B; higher
ambient temperatures were associated with higher vegetarian sales in College A but lower
vegetarian sales in College B. However, increasing vegetarian availability increased
vegetarian sales consistently in a similar way across colleges, indicating a strong and
potentially generalizable effect.
8
Vegetarian sales: individual-level data
1394 identifiable individual diners at College A and 746 at College B used the cafeteria
during the study period; this excludes guests and cash-only diners. Of these, 597 and 222
diners, respectively, purchased 10 meals in autumn 2016 (prior to our main study) and
were divided into quartiles within each college, based on their level of vegetarian meal
consumption during this period (Figure 1, Methods and SI Appendix Tables S7 and S8). In
both colleges every quartile from the Most Vegetarian to the Least Vegetarian bought more
vegetarian meals as vegetarian availability increased (Figure 2b&c). For both Colleges A and
B, the Least Vegetarian quartile had the strongest response to increasing vegetarian
availability (GLMM, College A, n= 32,687 meals, interaction effect size = 1.012 (CI= 1.004,
1.020), p=0.004; College B, n= 19,663 meals, interaction effect size= 1.024 (CI= 1.014,
1.034), p<0.001, SI Appendix Tables S9 and S10).
Total sales
College A sold an average of 191 main meals at a mealtime, and College B, 134. When
adjusted for other variables, increasing vegetarian availability had no significant effect on
total sales in College A and a small negative effect in College B where the mean total meals
sold decreased from 138 (CI= 129, 147) to 128 (CI= 118, 137) as vegetarian availability
increased from 25% to 50% (LM for main meals sold at a mealtime: College A, n=51,251
meals, availability effect size= 1.001 (CI= 0.997, 1.003), p=0.707; College B, n=35,681 meals,
availability effect size= 0.998 (CI= 0.997, 0.999), p<0.001)(Figure 2d and SI Appendix Tables
S11 and S12). The different quartiles of diners in College A did not respond differently, in
terms of number of meals bought at a mealtime, as vegetarian availability increased (LM,
n=33,180 meals, interaction terms p>0.05). In College B those in the Least Vegetarian
quartile responded more negatively to increasing vegetarian availability than those in other
quartiles, in terms of total number of meals purchased (LM, n=19,950 meals, interaction
effect size= 0.995 (CI= 0.992, 0.998), p<0.001). This was, however, still a small drop from a
mean of 27.4 (CI= 26.2%, 28.6%) meals to 24.7 (CI= 23.2%, 25.9%) as vegetarian availability
increased from 25% to 50%.
9
Figure 2: Effects of vegetarian availability on vegetarian and total sales for Study 1. a) Raw values (jittered) of
vegetarian sales against vegetarian availability; b and c): Modelled likelihood of selecting a vegetarian meal for
individual diners at Colleges A and B, with individual diners divided into Least Vegetarian to Most Vegetarian
quartiles; d) Raw values (jittered) of total sales against vegetarian availability. Lines of best fit and confidence
intervals generated from the models using conditional regression and the visreg package in R (Methods).
10
Study 2: experimental
Aims and design
We tested the causality of the association between vegetarian availability and vegetarian
sales by running an experiment at College C in autumn term 2017 based on fortnightly
alternation between one (control) and two (experiment) vegetarian options at lunchtimes
(Methods, SI Appendix Tables S13 and S14 and Figure S1). We analysed data from 44
lunchtimes. Vegetarian availability ranged from 16.7% to 50%, (impacted by differences in
the total number of options served, as well as our manipulation, SI Appendix Table S15).
Vegetarian sales: aggregate data
Vegetarian availability alone explained only 3.9% of the variation in vegetarian sales
(Binomial GLM, n=7712 meals, McFadden’s pseudo R2=0.039, p<0.001) in a univariate
analysis. When controlling for other variables (Methods) 31.8% of the variation was
explained (day of the week, week of term and the price differential of vegetarian and meat
meals were the predictors which explained most of the variation in vegetarian sales), and
availability remained a highly significant predictor of vegetarian sales (p<0.001, Figure 3a
and SI Appendix, SI Appendix Table S16). The model estimated that doubling vegetarian
availability from 25% to 50% increases vegetarian sales by 40.8% (from 19.1% (CI= 15.1%,
23.9%) to 26.9% (CI= 21.5%, 33.1%) of total sales, SI Appendix Table S16).
Vegetarian sales: individual-level data
121 of the 491 individual diners who bought a main meal during our experiment could be
assigned a quartile based on their level of vegetarian meal consumption in the previous
term, summer 2017 (Figure 1, SI Appendix Tables S17 and S18). When other variables were
controlled for, diners in every quartile (except Most Vegetarian) bought more vegetarian
meals in response to increasing vegetarian availability (SI Appendix Table S19). Similarly to
Study 1, for College C the Least Vegetarian quartile of diners had a significantly stronger
response to increasing vegetarian availability than the other quartiles (GLMM, n=1585
meals, interaction term effect size= 1.053 (CI= 1.002, 1.106), p=0.041, Figure 3b and SI
Appendix Table S19).
11
Total sales and possible rebound effects
College C sold an average of 175 meals per lunchtime and increasing vegetarian availability
had no effect on total sales (LM for main meals sold at lunchtime: n=7712 meals, availability
effect size= 1.000 (CI= 0.993, 1.004), p=0.942; Figure 3c and SI Appendix Table S20).
Moreover the different quartiles of diners responded similarly to each other in terms of
numbers of meals bought at a mealtime as vegetarian availability increased (LM, n=3201
meals, interaction terms p>0.1). In College C, unlike in Study 1, vegetarian sales at
dinnertimes could be used to explore possible rebound effects. We analysed dinner sales for
the 71% of autumn term lunchtime diners who also ate at dinner. When adjusted for other
variables, they bought similar numbers of vegetarian meals during the experimental weeks
(when there were two vegetarian options at lunchtimes) as in the control weeks (with one
vegetarian option)(GLM, control v experimental weeks, n=5287 meals, experimental weeks
effect size= 0.953 (CI= 0.795, 1.141), p=0.601, Figure 3d and SI Appendix Table S21). Hence
we found no evidence for a rebound effect involving a drop in vegetarian sales at
dinnertimes during weeks when there were higher vegetarian sales at lunchtimes.
12
Figure 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) Modelled likelihood of selecting a vegetarian meal for
individual diners, divided into Least Vegetarian to Most Vegetarian quartiles; 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 confidence intervals in white. Lines of best fit and confidence intervals
in a) and c) and model mean estimate with confidence intervals in d) generated from the models using
conditional regression and the visreg package in R (Methods).
13
Discussion
In all three participating colleges across Study 1 and Study 2 increasing the proportion of
vegetarian meals offered increased vegetarian sales, with a large effect size which was
greatest amongst those who prior to the study were less likely to select vegetarian meals.
To our knowledge this is the first year-long study on how altering availability affects
sustainable food choices. From 94,644 meals selected we found that doubling vegetarian
availability from 25% to 50% increased vegetarian sales (and decreased meat sales) by 7.8,
14.9 and 14.5 percentage points, equivalent to 40.8%, 61.8% and 78.8% increases.
Increasing vegetarian availability had little effect on total sales or vegetarian sales at other
mealtimes not involved in experiments, indicating rebound effects were probably small or
non-existent. In two out of three cafeterias increasing vegetarian availability did not to lead
different responses, in terms of number of meals bought, by diners with different prior
levels of vegetarian meal selection. In the third college there was a modest difference (with
those previously eating meat responding slightly negatively to increasing vegetarian meal
availability) but together these results suggest that increasing vegetarian availability did not
substantially put off meat eaters.
Although it might seem intuitive that providing proportionally more vegetarian options
would increase vegetarian sales, to our knowledge, this is untested. If meal preferences
were fixed, changing the availability of vegetarian options would have no effect. If meal
selections were random, this would lead to sales tracking the proportion of each meal
option available. Our results indicate that meal selection is neither fixed nor random but
rather is partially determined by availability. These results suggest that increasing the
proportion of vegetarian options may have a larger effect than many other choice
architecture interventions included in a recent systematic review on meat selection and
consumption (13): in previous studies neither restructuring food menus with different meal
descriptions nor positioning meat in less prominent positions reduced meat uptake.
Providing US and UK participants with meat substitutes, recipes and educational materials
led to large reductions in meat consumption (13): a 40% reduction in red and processed
(31), a 54% reduction in spending on meat (32), and a 70% reduction in meat consumed
(33). These results are impressive but, unlike increasing vegetarian availability, are time- and
14
resource-intensive so may not be scalable and their effects can diminish over time (24,
31): one paper found that at the end of the intervention meat consumption was 60% lower
than at the baseline but after two months the effect had decreased to 40% (31). Reducing
the serving size of meat portions reduced meat consumption by 13-14% (34, 35); hence
increasing vegetarian availability combined with smaller meat portions could be a powerful
combined strategy to reduce the mass of meat served by cafeterias.
Our studies have several strengths. While many recent papers have stressed the importance
of reducing meat consumption (13, 36) very few studies have tested which interventions
might work. For example, a recent systematic review found only 18 studies with 11,290
observations that tested how changing some aspect of choice architecture could reduce
meat consumption (13). Our studies have 94,644 observations from months of robust,
individual-level data. We collected both observational and experimental data and included
analyses on total meal sales. We have shown that increasing vegetarian availability can
substantially reduce meat consumption, even for those with low prior levels of vegetarian
meal consumption the most important demographic group to shift to reduce the GHGE of
the food system (5).
However, our studies also have several limitations. First, due to the design of the studies, we
did not collect data on the nutrition of the cafeteria meals or their palatability to students,
which are important considerations for catering managers (12, 37). Second, in keeping with
other similar field studies (22), some data were misclassified. Miscoding of a small number
of vegetarian meals as meat meals in College C led to a slight underestimate in Study 2 of
the effect of vegetarian availability on vegetarian sales (Methods), however this is highly
unlikely to change the results in a significant direction.
The current studies suggest opportunities for future research. First, they were conducted in
a university setting with students and staff. While this is a good context in which to generate
proof-of-concept evidence for the intervention, studies are now needed in other types of
food outlets, serving other populations including those in middle and low income countries
to estimate the generalisability of the current findings. Second, we were informed by
catering managers that ingredients costs were considerably cheaper for vegetarian meals,
15
but that labour costs might be higher. Future research could investigate the effects of
increasing sales of vegetarian meals on profits. Third, to achieve tangible environmental
benefits, any reduction in demand for meat needs to lead to reduced livestock farming, and
not simply redirecting livestock products to other countries (38). Shifting both diets and
agricultural production towards less meat will require the support of governments and
farmers as well as pressure from citizens (38, 39).
Nevertheless, our results demonstrate the potential of choice architecture for making
progress towards improved sustainability. Increasing the availability of vegetarian options in
cafeterias is a relatively cheap and easily-implemented strategy which generally goes
unnoticed: it does not require restructuring the canteen layout, or running meat-free days
that can prove unpopular (40), and it can save money on ingredients (24). Increasing the
availability of plant-based meals will require diversification of vegetarian provision by
cafeterias and restaurants which may in turn necessitate changes in the training offered to
chefs (37). Interest in reducing meat consumption and in “flexitarianism” is on the rise (41)
and our results show that caterers serving more plant-based options are not just responding
to but also re-shaping customer demand. Further long-term studies intervening on
availability in addition to other aspects of choice environments, and conducted in a wider
range of settings might usefully test behavioural interventions that are scalable and offer
the potential to significantly mitigate climate change and biodiversity loss.
16
Methods
This research was approved by the University of Cambridge Psychology Research Ethics
Committee (PRE.2016.100). In keeping with research governance for interventions that
target environments and not individuals directly, consent was obtained from those who
have authority over these environments, i.e. the managers of the college cafeterias. Signed
consent forms, approved by the Research Ethics Committee, were obtained from each of
the catering managers of the three participating colleges.
Study setting
Colleges A and B have both undergraduate and postgraduate members. College A has over
1100 members, and College B over 500. College C is a graduate college with over 600
members. All three colleges admit students of any gender identity. Students pay for meals
by swiping their university cards, meals are not included in the tuition or accommodation
fees. In Colleges A and B, students top up their card with credit throughout the academic
year, in College C students pay the bill at the end of each term. Meals typically cost between
£2.30 [€2.51, $2.45] and £3.70 [€4.04, $4.50]. Although many students eat in the college
cafeteria, others cook their own meals or eat elsewhere. In the cafeterias vegetarian and
meat meals are available throughout the mealtime, if meat or vegetarian options run out
they are quickly replaced by an option in the same category.
Study design
Study 1
Colleges A and B in their normal operations varied both the total number of options and the
number of vegetarian options available. We did not experimentally alter the menus from
these colleges but observed how the availability of vegetarian meals related to their relative
sales. We used data from lunch and dinner on weekdays (Monday to Friday) during spring
(16th January to 17th March), summer (24th April to 30th June) and autumn terms (2nd
October to 1st December) 2017.
17
Study 2
College C experimentally altered the number of vegetarian meals on their menus. The
original experimental design specified that that both lunch and dinner would alternate
between one and two vegetarian options week by week. However, this was too much for
the cafeteria to implement within the timeframe of the study. Therefore, only lunchtimes
alternated between the experimental condition of one and two vegetarian options, every
two weeks. The number of vegetarian options still sometimes varied from experimental
allocation due to cafeteria constraints (SI Appendix Table S15). Some misclassifications at
the checkout occurred, resulting in some vegetarian meals being recorded as meat sales.
This meant that vegetarian sales may have been up to 21.5% greater than recorded (EG,
pers. obs.). No meat meals were misclassified as vegetarian. Though unfortunate, this error
is conservative and suggests that the true effect of availability at College C could be
substantially greater than that reported, and closer to that estimated from the
observational work at Colleges A and B.
We collected and analysed the experimental data from weekday lunchtimes from College C
to test the effect of vegetarian availability, and also compared this with weekday dinner
sales to investigate if increasing vegetarian availability at lunch affected vegetarian sales at
dinner. Data were collected across autumn term and the first two weeks of the Christmas
holidays 2017 (2nd October to 15th December). Unlike College A and B, College C is a
graduate college and meals were served to staff and students outside of normal university
term-times, so to increase the sample size we included the first two weeks of the Christmas
holidays. These two weeks did have slightly lower total sales than term time weeks (SI
Appendix Table S19) but did not have significantly different vegetarian sales (SI Appendix
Table S15).
Data collection
Sales data were downloaded from the online catering platforms Uniware (42) and Accurate
Solutions (43) and identifiable data were stored on a secure online server. All three colleges
had online menus; however the options served sometimes varied from this. At Colleges A
and B the number of vegetarian options and total number of options could be inferred from
how the sales data are coded. At College C it was not possible to infer the number of
vegetarian options and total options from the sales data, therefore visits were made at
18
lunchtimes to directly observe the options available. When the lunch offer included a pasta
bar this commonly had two sauces, often one vegetarian and one meat; we counted each
sauce+pasta as half an option.
Data preparation
We summarised the sales data into a) aggregate data, summarising the total vegetarian and
meat sales at each lunch and dinner and b) individual-level data on whether each individual
diner at a meal selected a vegetarian or meat meal. Eight mealtimes at College A and three
at College B served no vegetarian main meals, and therefore vegetarian availability and
vegetarian sales were zero. These data were excluded from the analysis to avoid
overestimating the effect of availability (SI Appendix Table S3). In College B one mealtime
only served one main meal in total and this was also excluded from the analysis. Only
lunchtimes when direct observations were made of the vegetarian and total options
available were included in the analysis for College C.
Aggregate data included main meals bought by both college members and guests.
Individual-level data only included meals bought by college members on their university
cards, as only these meals could be associated with individual diners. An individual diner
who bought one or more vegetarian meals at a mealtime was coded as 1; an individual diner
who bought one or more meat meals was coded as 0. Any individual diners who bought
both vegetarian and meat meals at one meal time were coded as NA and we excluded those
meal choices from the analysis; this removed 1.6% of the individual-level data at College A
(699/43,751), 1.5% at College B (468/31,956) and 4.5% at College C (207/4,565).
We wanted to test if the response to increasing vegetarian availability varied with
background levels of meat consumption. To calculate this, for individuals who bought 10
main meals during the preceding term (autumn 2016 for Colleges A and B, summer term
2017 for College C), we calculated the proportion of main meals bought that were
vegetarian, and these values were used to divide the individual diners into within-college
quartiles: Least, Less, More and Most Vegetarian.
19
Statistical approaches
We carried out analyses in R 3.5 (44), using the lme4 (45) packages. We used Binomial
Generalised Linear Models for the aggregate data, and Binomial Generalised Linear Mixed
Models for the individual-level data with each individual diner included as a random effect.
Models were evaluated using AIC values and interpretability. We follow the
recommendations of Simmons et al (46), which includes citing the effect of vegetarian
availability, with and without covariates. Initial analyses showed that relative vegetarian
availability (number of vegetarian options/ number of total options) was a better predictor
of vegetarian sales than number of vegetarian or meat options and therefore we used this
as the predictor variable for vegetarian availability. We estimated the effect of vegetarian
availability on vegetarian sales and total sales, adjusting for other pre-determined variables
(Table 1). After model selection, we used the predict function to generate the predicted
values and plotted out lines of best fit, using conditional regressions with 95% confidence
intervals using the effects (47) and visreg packages (48).
Table 1: Variables considered for statistical models.
Model
Variable
Description
All models
Vegetarian
availability
Number of vegetarian options/ total options available
Total options
available
Number of different meal options offered at a
mealtime
Total main meals
sold
Number of main meals sold at a mealtime
Vegetarian price
differential (£)
The difference between the mean cost of the meat
options and the vegetarian options
Ambient
temperature
(centigrade)
Mean temperature over 24 hours each day in
Cambridge(49)
Day
Monday, Tuesday, Wednesday, Thursday, Friday
Week of term
1-11
For Study 1 only (no variation
in Study 2)
Meal
Lunch or dinner
Term
Spring, summer, autumn
For individual-level models
only
Individual diner as a
random effect
For individual-level models
and models of total sales
considering diner background
Prior level of
vegetarian meal
consumption
Individual diners at each college were divided into
Least, Less, More and Most Vegetarian quartiles and
we tested for any interaction effects with vegetarian
availability
For Study 2 rebound model
Week condition
Control or experimental week
20
Acknowledgements
We thank Ivan, Paul, Rob, Alex, Gary and Christine for generously contributing their time
and data, and Benno Simmons for help with coding in R. E.G is funded by a NERC
studentship, grant number NE/L002507/1. A.B. is supported by a Royal Society Wolfson
Research Merit award.
Author contributions
E.G., A.B., C.S. and T.M. designed research; E.G. performed research; E.G. and M.P. analysed
data; E.G., A.B., C.S. and T.M. wrote the paper.
Conflict of interest
The authors declare no conflict of interest.
21
References
1. Bryngelsson D, Wirsenius S, Hedenus F, Sonesson U (2016) How can the EU climate targets be met? A
combined analysis of technological and demand-side changes in food and agriculture. Food Policy
59:152164.
2. Poore J, Nemecek T (2018) Reducing food’s environmental impacts through producers and consumers.
Science (80- ) 992(6392):987992.
3. Clune S, Crossin E, Verghese K (2017) Systematic review of greenhouse gas emissions for different
fresh food categories. J Clean Prod 140:766783.
4. Springmann M, et al. (2017) Mitigation potential and global health impacts from emissions pricing of
food commodities. Nat Clim Chang (7):6974.
5. Scarborough P, et al. (2014) Dietary greenhouse gas emissions of meat-eaters, fish-eaters, vegetarians
and vegans in the UK. Clim Change 125(2):179192.
6. Reddy SMW, et al. (2016) Advancing Conservation by Understanding and Influencing Human Behavior.
Conserv Lett 00(May):19.
7. Marteau TM (2017) Towards environmentally sustainable human behaviour: targeting non-conscious
and conscious processes for effective and acceptable policies. Philos Trans R Soc A Math Phys Eng Sci
375(i). doi:10.1098/rsta.2016.0371.
8. Diepeveen S, Ling T, Suhrcke M, Roland M, Marteau TM (2013) Public acceptability of government
intervention to change health-related behaviours: a systematic review and narrative synthesis. BMC
Public Health 13(1):756.
9. Bianchi F, Dorsel C, Garnett E, Aveyard P, Jebb SA (2018) Interventions targeting conscious
determinants of human behaviour to reduce the demand for meat: A systematic review with
qualitative comparative analysis. Int J Behav Nutr Phys Act 15(102):125.
10. Bucher T, et al. (2016) Nudging consumers towards healthier choices: a systematic review of positional
influences on food choice. Br J Nutr (9):112.
11. Cadario R, Chandon P (2017) Which Healthy Eating Nudges Work Best? A Meta-Analysis of Behavioral
Interventions in Field Experiments.
12. Campbell-Arvai V, Arvai J, Kalof L (2014) Motivating sustainable food choices: The role of nudges, value
orientation, and information provision. Environ Behav 46(4):453475.
13. Bianchi F, Garnett E, Dorsel C, Aveyard P, Jebb SA (2018) Restructuring physical micro-environments to
reduce the demand for meat: A systematic review with qualitative comparative analysis. Lancet Planet
Heal 2(9):e384e397.
14. McKinsey Global Institute (2014) Overcoming obesity: An initial economic analysis.
15. Hollands GJ, et al. Altering the availability or proximity of food, alcohol and tobacco products to change
their selection and consumption. Cochrane database Syst Rev (3).
doi:10.1002/14651858.CD011672.READS.
16. Fiske A, Cullen KW (2004) Effects of promotional materials on vending sales of low-fat items in
teachers’ lounges. J Am Diet Assoc 104(1):9093.
17. Foster GD, et al. (2014) Placement and promotion strategies to increase sales of healthier products in
supermarkets in low-income, ethnically diverse neighborhoods: A randomized controlled trial. Am J
Clin Nutr 99(6):13591368.
22
18. Roe LS, Meengs JS, Birch LL, Rolls BJ (2013) Serving a variety of vegetables and fruit as a snack
increased intake in preschool children. Am J Clin Nutr 98(3):693699.
19. Stubbs RJ, Johnstone AM, Mazlan N, Mbaiwa SE, Ferris S (2001) Effect of altering the variety of
sensorially distinct foods, of the same macronutrient content, on food intake and body weight in men.
Eur J Clin Nutr 55(1):1928.
20. Kocken PL, et al. (2012) Promoting the Purchase of Low-Calorie Foods From School Vending Machines:
A Cluster-Randomized Controlled Study. J Sch Health 82(3):115122.
21. Bartholomew JB, Jowers EM (2006) Increasing frequency of lower-fat entrees offered at school lunch:
An environmental change strategy to increase healthful selections. J Am Diet Assoc 106(2):248252.
22. Pechey R, et al. (2019) Impact of increasing the proportion of healthier foods available on energy
purchased in worksite cafeterias: A stepped wedge randomized controlled pilot trial. Appetite
133(October 2018):286296.
23. Clark M (2017) Chronic effects of replacing red and processed meat with non/reduced meat
alternatives. Dissertation (University of Cambridge).
24. Gravert C, Kurz V (2017) Nudging À La Carte A Field Experiment on Food Choice
doi:10.2139/ssrn.2909700.
25. Bacon L, Krpan D (2018) (Not) Eating for the environment: The impact of restaurant menu design on
vegetarian food choice. Appetite 125:190200.
26. Whitaker RC, Wright JA, Finch AJ, Psaty BM (1993) An environmental intervention to reduce dietary fat
in school lunches. Pediatrics 91(6):11071111.
27. Grech A, Allman-Farinelli M (2015) A systematic literature review of nutrition interventions in vending
machines that encourage consumers to make healthier choices. Obes Rev 16(12):10301041.
28. O’Reilly GA, et al. (2017) Sugar Restriction Leads to Increased Ad Libitum Sugar Intake by Overweight
Adolescents in an Experimental Test Meal Setting. J Acad Nutr Diet 117(7):10411048.
29. McCulloch CE, Searle SR, Neuhaus JM (2008) Generalized, Linear, and Mixed Models (Wiley-Blackwell).
Second Edi.
30. Hosmer DW, Lemeshow S, Sturdivant RX (2013) Applied Logistic Regression. Applied Logistic
Regression (Wiley), pp 89152. Third Edit.
31. Ali M, Simpson EJ, Clark M, Razak A, Salter A (2017) The impact of dietary meat intake reduction on
haematological parameters in healthy adults. Proc Nutr Soc 76(OCE3):E70.
32. Flynn MM, Reinert S, Schiff AR (2013) A Six-Week Cooking Program of Plant-Based Recipes Improves
Food Security, Body Weight, and Food Purchases for Food Pantry Clients. J Hunger Environ Nutr
8(1):7384.
33. Holloway T, Salter AM, Mccullough FS (2012) Chapter 4: Dietary intervention to reduce meat intake by
50 % in University students a pilot study. Proc Nutr Soc 71(OCE2):119151.
34. Reinders MJ, Huitink M, Dijkstra SC, Maaskant AJ, Heijnen J (2017) Menu-engineering in restaurants -
adapting portion sizes on plates to enhance vegetable consumption: a real-life experiment. Int J Behav
Nutr Phys Act 14(1):41.
35. Rolls BJ, Roe LS, Meengs JS (2010) Portion size can be used strategically to increase vegetable
consumption in adults. Am J Clin Nutr 91(4):913922.
36. Godfray HCJ, et al. (2018) Meat consumption, health, and the environment. Science (80- ) 243(July).
doi:10.1126/science.aam5324.
23
37. Volkhardt I, et al. (2016) Checklist for a vegan lunch menu in public catering. Ernaehrungs Umschau
63(09):176184.
38. Buckwell A, Nadeu E (2018) What is the Safe Operating Space for EU livestock? (Brussels) Available at:
www.risefoundation.eu/publications.
39. Committee on Climate Change (2018) Land use: Reducing emissions and preparing for climate change.
40. Lombardini C, Lankoski L (2013) Forced Choice Restriction in Promoting Sustainable Food
Consumption: Intended and Unintended Effects of the Mandatory Vegetarian Day in Helsinki Schools. J
Consum Policy 36(2):159178.
41. Eating Better (2017) The future of eating is flexitarian: Companies leading the way.
42. Uniware Available at: http://www.uniware.co.uk/ [Accessed January 8, 2019].
43. Accurate Solutions Available at: http://www.accurate-solutions.co.uk/ [Accessed January 8, 2019].
44. R Core Team (2018) R: A language and environment for statistical computing. Available at:
https://www.r-project.org/.
45. Bates D, Bolker BM, Maechler M, Walker SC (2015) Fitting Linear Mixed Effects Models using lme4. J
Stat Softw 67(1):251264.
46. Simmons JP, Nelson LD, Simonsohn U (2011) False-Positive Psychology: Undisclosed Flexibility in Data
Collection and Analysis Allows Presenting Anything as Significant. Psychol Sci 22(11):13591366.
47. Fox J, Weisberg S (2018) Visualizing Fit and Lack of Fit in Complex Regression Models with Predictor
Effect Plots and Partial Residuals. J Stat Softw 87(9):127.
48. Breheny P, Burchett W (2016) Visualization of Regression Models Using visreg. R J 9(December):5671.
49. Cambridge Daily Weather Graphs (2018) Available at:
http://www.cl.cam.ac.uk/research/dtg/weather/index-daily-graph.html.
... The principal aim of this study was to investigate how customers with different lunchtime eating habits concerning their preference for meat responded to a revised choice of meat and lacto-vegetarian or vegan dishes (veg-dishes). To the authors' knowledge, there are hardly any studies that have analysed the revealed preferences for meat dishes based on real POS data [28]. To explore this research question, data from the field experiment in two Swiss university canteens in the autumn semester of 2017 were used. ...
... Concerning the relative increase of veg-dishes, meat-oriented customers reacted more to the revised menu choice: 'Meat lover' more than doubled the share of veg-dishes, but from a lower starting point. This result corresponds to a study by Garnett et al. (2019) [28], where diners also responded to a higher availability of vegetarian dishes, with the largest effects in the quartile of participants who were previously least likely to buy a vegetarian meal. ...
... Concerning the relative increase of veg-dishes, meat-oriented customers reacted more to the revised menu choice: 'Meat lover' more than doubled the share of veg-dishes, but from a lower starting point. This result corresponds to a study by Garnett et al. (2019) [28], where diners also responded to a higher availability of vegetarian dishes, with the largest effects in the quartile of participants who were previously least likely to buy a vegetarian meal. ...
Article
Full-text available
In a transdisciplinary field experiment in two Swiss university canteens over 12 weeks, how customers respond to a revised menu choice of meat dishes and ovo-lacto-vegetarian or vegan dishes was investigated. Several interventions were implemented: the vegetarian menu line was abolished, the meat and veg-dishes were randomly distributed across the three menu lines, vegetarian and vegan dishes (veg-dishes) were not labelled or marketed as such, and in the 6 ‘intervention weeks’ the share of veg-dishes was increased compared to the 6 ‘base weeks’. Based on the lunchtime eating habits in the canteen, regular canteen visitors were categorised into either seven or five different ‘lunch types’. During the intervention weeks, the share of meat dishes decreased for every lunch type, although with different response patterns. It is particularly noteworthy that customers with meat-oriented lunchtime eating habits ate in the canteen more frequently. This indicates that the usual choices in the two canteens are more appealing for meat-eaters. Overall, the data show a clear potential to reduce meat consumption in canteens by increasing the range of attractive veg-dishes without explicitly marketing them as vegetarian or vegan. We conclude this also to be an economically profitable strategy, as it provides a chance to attract customers with veg-oriented lunchtime eating habits without either losing meat eaters or meat lovers.
... For binary meal choice indicators, we also control for the number of options available for the dependent variable in question and the average price differential between high and low-carbon alternatives. Both availability and price have been found to play an important role in food purchase decisions (Garnett et al., 2021;Garnett et al., 2019). 15 = × is an indicator for a treated individual (Treat) during the labelling period (Post), with both main effects being subsumed by the individual and week fixed effects. ...
... Tables 4.3.6 and 4.3.7 show the estimated average treatment effect of the carbon footprint labels on climatarian meal choices for each sub-sample of vegetarian and climatarian preferences, respectively, estimated by OLS using the generalised DID specification shown in (4.1). 20 Following Garnett et al. (2019), we selected 10 observations as the minimum number of cafeteria visits to accurately classify individual dietary preferences. While setting a higher threshold would further increase accuracy, this would result in a smaller sample size by restricting the sample to the more frequent customers. ...
Thesis
Human behaviour lies at the heart of the climate crisis. Not only is it the primary cause of global climate change and environmental degradation, but also key to responding and adapting to them. Tackling the climate crisis thus requires a complete understanding of human behaviour. So far, however, environmental policies have largely been guided by the canonical economic model of human behaviour, based on the idea of ‘homo economicus’, neglecting important behavioural aspects of the relationship between human behaviour and the environment. This thesis examines some of the complex interrelations between human behaviour and the environment through a behavioural environmental economics lens, drawing on recent insights from behavioural economics and psychology. The first part of this thesis (Chapters 1 and 2) focuses on the impact of environmental stressors on human behaviour, attitudes and beliefs. The second part (Chapters 3 and 4) examines the impact of policy interventions to foster more environmentally sustainable behaviour. All chapters adopt an experimental or quasi-experimental approach to provide causal insights and formulate robust policy recommendations. Chapter 1 develops and tests a novel experimental design, that exploits natural discontinuities in air pollution episodes in Beijing, China to experimentally isolate the causal effect of acute air pollution on social decision-making and economic preferences. Chapter 2 utilises data from a natural experiment to estimate the causal effect of extreme weather events on climate change attitudes and pro-environmental behaviours. Chapter 3 uses an online ‘message framing’ experiment to explore whether appealing to ‘warm glow’ motives can encourage voluntary pro-environmental behaviour, relative to other common climate change communication strategies. Chapter 4 presents the results of a large-scale field experiment conducted at five university cafeterias, exploring whether carbon footprint labels can promote more sustainable food choices.
... Tunisia (61.2%), and Sierra Leone (50.0%), where plant-derived AAs are the major source of EAAs which also causes significant reduction of greenhouse gases in the environment (DellaPenna 1999;Millward 2016;Garnett et al. 2019). Therefore, an increase in plant EAA content could replenish EAAs in large populations suffering from malnutrition. ...
Article
Full-text available
Ruminants and humans are unable to synthesize essential amino acids (EAAs) and conditionally essential amino acids (CEAAs) under normal conditions and need to acquire them from plant sources. Maize plays, as a major crop, a central role in global food security. However, maize is deficient in several EAAs and CEAAs. Genetic engineering has been successfully used to enrich the EAA content of maize to some extent, including the content of Lys, Trp, and Met. However, research on other EAAs is lacking. Genetic engineering provides several viable approaches for increasing the EAA content in maize, including transformation of a single gene, transformation of multiple genes in a single cassette, overex-pression of putative amino acid transporters, engineering the amino acid biosynthesis pathway including silencing of feedback inhibition enzymes, and overexpression of major enzymes in this pathway. These challenging processes require a deep understanding of the biosynthetic and metabolic pathways of individual amino acids, and the interaction of individual amino acids with other metabolic pathways.
... Behavior intervention strategies promise to be effective tools to nudge consumers toward plant-based food choices. A study performed in three university cafeterias, for example, showed that doubling the proportion of vegetarian meals offered increased sales of between 41% and 79%, indicating that food availability and substitution may be an effective strategy for increasing consumption of plant-based meals (Garnett et al., 2019). Another study measuring food choices in a university cafeteria demonstrated that taste-focused labeling of vegetables in dining halls increased vegetable consumption by 39%, compared to health-centered labeling of the same dish (Turnwald et al., 2017), adding to a growing body of evidence indicating that menu design and labeling can influence consumer acceptance (Vadiveloo et al., 2017). ...
Article
Full-text available
Evidence consistently suggests that plant-based diets promote human and planetary health. Reducing large-scale animal-based food production generates environmental benefits, as the entire livestock agriculture chain plays an outsized role in greenhouse gas emissions, land change and degradation, and scarcity-weighted water use. However, substituting animal products with their plant-based counterparts must come with consideration of the nutritional quality and resource usage of plant-based food production and processing operations. Several policy reforms have been implemented at the national, state, and municipal levels in the United States to support a transition toward more plant-based diets. Federal programs such as the Supplemental Nutrition Assistance Program and the Dietary Guidelines for Americans generally promote the consumption of unprocessed plant-based foods but include little to no information on sustainability and the harmful environmental impact of animal-based foods. National policies are complemented by state efforts aimed at incentivizing produce purchased from local suppliers and encouraging resource-conserving agriculture. At the local level, public schools are implementing programs to promote plant-based protein on their menus, and urban gardens are sprouting across the country to increase access to organic farming. This mini-review examines these policy reforms and behavioral intervention strategies, based on the social-ecological model, and discuss their capacity and limitations to promote a shift toward sustainably produced plant-based diets in the United States. We conclude that transforming the food systems toward plant-based diets in the animal-centered United States requires multi-sector collaboration and context-specific policy solutions to address diet-related climate concerns without neglecting health, social, and financial constraints.
... This is also consistent with field studies suggesting that increasing the availability of lower-energy (e.g. [1,2]) and plant-based [3] foods increases their selection. The absolute levels of lower-energy option selection varied between Study 1 and 2, with 21% and 12% respectively selecting a lower-energy option when there were predominantly higher-energy options, which approximately tripled when there were predominantly lower-energy ones. ...
Article
Full-text available
Background Increasing the availability of healthier or plant-based foods increases their selection. The current studies aimed to examine the extent to which relative preferences account for food selections following availability interventions. In particular, (a) whether increasing the availability of lower-energy options increases the likelihood that individuals’ highest-ranked option is lower-energy, and (b) the extent to which selections reflect individuals’ highest-ranked option from the available range. Methods UK adults (Study 1: n = 1976; Study 2: n = 1078) took part in within-subjects online studies. In both studies, the order of preference between food options was established by participants choosing the option that they would prefer “to eat right now” from every possible pairing within a pool of eight options. Then, participants were shown either predominantly higher-energy options (three higher- and one lower-energy) or predominantly lower-energy options (vice versa), presented in a random order. Results When predominantly lower-energy options were presented, the odds of the highest-ranked option being a lower-energy option increased ten-fold (Study 1: odds ratio: 10.1; 95%CI: 8.9,11.4; Study 2: odds ratio: 10.4; 95%CI: 7.4,14.7), compared to when predominantly higher-energy options were available. In both studies, around 90% of selections reflected the highest-ranked option in the range offered in the studied availability conditions (range 88–92%). Conclusions These studies suggest that increased availability of lower-energy options increases the likelihood of an individual’s highest-ranked option being lower-energy, and that the highest-ranked option has the greatest likelihood of selection. As such, preferences may be a key contributor to the effects of altering availability on food selections. Trial registration ISRCTN ( http://www.isrctn.com/ISRCTN27598623 ; 3/12/19 [Study 1]; http://www.isrctn.com/ISRCTN61010183 ; 20/4/20 [Study 2]).
... Other interventions rely on unconscious behavioural responses to contextual cues. For instance, increasing the proportion of vegetarian meals offered at cafeterias, 229 and programming restaurant menus and online meal registrations 230 by offering vegetarian dishes as a default option, can increase the number of people choosing them. ...
Article
Full-text available
A debate has emerged over the potential socio-ecological drivers of wildlife-origin zoonotic disease outbreaks and emerging infectious disease (EID) events. This Review explores the extent to which the incidence of wildlife-origin infectious disease outbreaks, which are likely to include devastating pandemics like HIV/AIDS and COVID-19, may be linked to excessive and increasing rates of tropical deforestation for agricultural food production and wild meat hunting and trade, which are further related to contemporary ecological crises such as global warming and mass species extinction. Here we explore a set of precautionary responses to wildlife-origin zoonosis threat, including: (a) limiting human encroachment into tropical wildlands by promoting a global transition to diets low in livestock source foods; (b) containing tropical wild meat hunting and trade by curbing urban wild meat demand, while securing access for indigenous people and local communities in remote subsistence areas; and (c) improving biosecurity and other strategies to break zoonosis transmission pathways at the wildlife-human interface and along animal source food supply chains.
Article
Full-text available
Restaurants are characterized by high levels of meat being consumed in this out-of-home setting, while plant-based meat alternatives remain a niche product, thus preserving a high environmental impact of food consumption. We tested whether subtly re-designing the restaurant menu, so that plant-based meat alternatives were perceived as the default to a greater extent, increased consumer selection of plant-based meat alternatives. Consumers' freedom of choice was preserved by leaving all choice options on the menu. An online experiment in The Netherlands showed that consumers choose plant-based meat alternatives more often relative to meat when the plant-based option is framed as the default. In a field experiment in a Dutch restaurant, we found that the amount of ordered plant-based meat alternative dishes increased relative to an equivalent meat dish when implementing a default nudge (bean alternative: from 8.6% to 80.0%; seaweed alternative: from 16.1% to 58.3%). Thus, re-designing the menu in a way that suggests that plant-based meat alternatives are the default, while preserving autonomous decision-making, is a promising route to promote out-of-home adoption of plant-based meat alternatives in restaurants.
Article
Achieving sustainable healthy diets for all will require substantial changes in individuals' food choices. Motivation plays a central role in behaviour change, yet little is known about the relationship between food choice motives and sustainability of the overall diet. The present study explored the relationship between food choice motives and different facets of diet sustainability in order to emphasize individuals' motivational supports and barriers regarding diet sustainability. We analysed food choice motives and dietary data collected through online questionnaires in 938 French adults (79% female, mean age = 39y (SD = 12)). The importance of nine food choice motives was assessed and three overarching motivational dimensions were derived from a principal component analysis on food choice motives: ‘health and sustainability, ‘ease and accessibility’ and ‘pleasure’. Food intakes were recorded using a food frequency questionnaire and six indicators of diet sustainability were calculated: nutritional quality, environmental impact, adherence to the EAT-Lancet diet, organic food consumption, local food consumption and ready-to-eat meals consumption. ‘Health and sustainability’ motives were consistently positively associated with diet sustainability indicators. On the contrary, ‘ease and accessibility’ motives were consistently negatively associated with diet sustainability. Associations between ‘pleasure’ motives and diet sustainability were mixed. Based on these results, it seems equally important to promote individuals' awareness of health and environmental impacts of food as to increase the accessibility of convenient sustainable healthy food products to support individuals' food choices towards sustainable diets.
Article
Full-text available
Predictor effect displays, introduced in this article, visualize the response surface of complex regression models by averaging and conditioning, producing a sequence of 2D line graphs, one graph or set of graphs for each predictor in the regression problem. Partial residual plots visualize lack of fit, traditionally in relatively simple additive regression models. We combine partial residuals with effect displays to visualize both fit and lack of fit simultaneously in complex regression models, plotting residuals from a model around 2D slices of the fitted response surface. Employing fundamental results on partial residual plots along with examples for both real and contrived data, we discuss and illustrate both the strengths and limitations of the resulting graphs. The methods described in this paper are implemented in the effects package for R.
Article
Full-text available
Increasing the proportion of healthier foods available could encourage healthier consumption, but evidence to date is limited in scope and quality. The current study aimed to: (a) examine the feasibility and acceptability of intervening to change product availability in worksite cafeterias; and (b) estimate the impact on energy purchased of increasing the proportion of healthier (i.e. lower energy) cooked meals, snacks, cold drinks and sandwiches. Six English worksite cafeterias increased the proportion of healthier foods available, aiming to keep the total number of options constant, in a stepped wedge randomized controlled pilot trial conducted between January and May 2017. The intervention was generally successfully implemented and acceptable to clientele. Generalized linear mixed models showed a reduction of 6.9% (95%CI: -11.7%, −1.7%, p = 0.044) in energy (kcal) purchased from targeted food categories across all sites. However, impact varied across sites, with energy purchased from targeted categories significantly reduced in two sites (−10.7% (95%CI: -18.1% to −2.6%, p = 0.046); −18.4% (95%CI: -26.9% to −8.8%, p = 0.013)), while no significant differences were seen in the other four sites. Overall, increasing the proportion of healthier options available in worksite cafeterias seems a promising intervention to reduce energy purchased but contextual effects merit further study.
Article
Full-text available
Background Reducing meat consumption can help prevent non-communicable diseases and protect the environment. Interventions targeting conscious determinants of human behaviour are generally acceptable approaches to promote dietary change, but little is known about their effectiveness to reduce the demand for meat. Objective To evaluate the effectiveness of interventions targeting conscious determinants of human behaviour to reduce the demand for meat. Methods We searched six electronic databases on the 31st of August 2017 with a predefined algorithm, screened publicly accessible resources, contacted authors, and conducted forward and backward reference searches. Eligible studies employed experimental designs to evaluate interventions targeting conscious determinants of human behaviour to reduce the consumption, purchase, or selection of meat in comparison to a control condition, a baseline period, or relative to other eligible interventions. We synthesised results narratively and conducted an exploratory crisp-set Qualitative Comparative Analysis to identify combinations of intervention characteristics associated with significant reductions in the demand for meat. Results We included 24 papers reporting on 59 interventions and 25,477 observations. Self-monitoring interventions and individual lifestyle counselling led to, or were associated with reduced meat consumption. Providing information about the health or environmental consequences of eating meat was associated with reduced intentions to consume and select meat in virtual environments, but there was no evidence to suggest this approach influenced actual behaviour. Education about the animal welfare consequences of eating meat was associated with reduced intentions to consume meat, while interventions implicitly highlighting animal suffering were not. Education on multiple consequences of eating meat led to mixed results. Tailored education was not found to reduce actual or intended meat consumption, though few studies assessed this approach. Conclusion Some interventions targeting conscious determinants of human behaviour have the potential to reduce the demand for meat. In particular, self-monitoring interventions and individual lifestyle counselling can help to reduce meat consumption. There was evidence of effectiveness of some educational messages in reducing intended consumption and selection of meat in virtual environments. Protocol registration CRD42017076720. Electronic supplementary material The online version of this article (10.1186/s12966-018-0729-6) contains supplementary material, which is available to authorized users.
Technical Report
Full-text available
This report, co-authored by Allan Buckwell and Elisabet Nadeu, provides a comprehensive overview of the benefits and impacts of livestock on the environment and human health and calls upon policy makers to use the range of policy tools at their disposal to support the livestock sector through a transition that is considered necessary and inevitable.
Article
Full-text available
Background: Reducing meat consumption could help to protect the natural environment and promote population health. Interventions restructuring physical micro-environments might help to change habitual behaviour. We synthesised the scientific evidence pertaining to whether, and which, interventions restructuring physical micro-environments effectively reduce the demand for meat. Methods: We did a systematic review of quantitative studies evaluating the effectiveness of interventions restructuring physical micro-environments to reduce the demand for meat. We identified relevant records by searching six electronic databases (CAB Abstracts, Embase, PsycINFO, Science Citation Index, MEDLINE, and Dissertations & Theses) on Aug 31, 2017, contacting experts, screening publicly accessible online resources, and searching references. We included studies that evaluated the effectiveness of interventions restructuring physical micro-environments to reduce the demand for meat, defined as the actual or intended consumption, purchase, or selection of meat in real or virtual environments. We extracted data pertaining to the study samples, the interventions, and meat demand at the follow-up closest to intervention completion and at the longest follow-up, with the former representing our primary outcome. We synthesised data narratively and did a qualitative comparative analysis to identify configurations of intervention characteristics associated with, and those not found to be associated with, significant reductions in meat demand. Our Systematic Review is registered with PROSPERO, number CRD42017081532. Results: Of 10 733 titles and abstracts screened for eligibility, we assessed 60 full papers and included 14 papers reporting on 18 studies with 22 intervention conditions. Three interventions reducing the portion size of meat servings reduced meat consumption in randomised trials. Three interventions providing meat alternatives with supporting educational material were associated with reduced meat demand in pre-post design studies. Three of four interventions altering the sensory properties (eg, visual presentation) of meat or meat alternatives at point of purchase reduced meat demand in randomised trials. Four interventions repositioning meat products to be less prominent at point of purchase were associated with lower meat demand, but only two such interventions reached statistical significance in a randomised trial and a multiple treatment reversal design. Only one of five interventions manipulating the description of meat or meat alternatives at point of purchase was associated with lower meat demand in a multiple treatment reversal design. Evidence from randomised trials evaluating a pricing intervention or interventions restructuring several aspects of micro-environments was too scarce or inconsistent to be conclusive. The results from our qualitative comparative analysis supported the findings of this narrative synthesis. Interpretation: Some interventions restructuring physical micro-environments could help to promote lower demand for meat. Interventions reducing portion sizes of meat servings, providing meat alternatives, or changing the sensory properties of meat and meat alternatives at point of purchase offered the most promise in the context of experimental studies. Funding: None.
Article
Full-text available
Both the global average per capita consumption of meat and the total amount of meat consumed are rising, driven by increasing average individual incomes and by population growth. The consumption of different types of meat and meat products has substantial effects on people’s health, and livestock production can have major negative effects on the environment. Here, we explore the evidence base for these assertions and the options policy-makers have should they wish to intervene to affect population meat consumption. We highlight where more research is required and the great importance of integrating insights from the natural and social sciences.
Article
Full-text available
Food's environmental impacts are created by millions of diverse producers. To identify solutions that are effective under this heterogeneity, we consolidated data covering five environmental indicators; 38,700 farms; and 1600 processors, packaging types, and retailers. Impact can vary 50-fold among producers of the same product, creating substantial mitigation opportunities. However, mitigation is complicated by trade-offs, multiple ways for producers to achieve low impacts, and interactions throughout the supply chain. Producers have limits on how far they can reduce impacts. Most strikingly, impacts of the lowest-impact animal products typically exceed those of vegetable substitutes, providing new evidence for the importance of dietary change. Cumulatively, our findings support an approach where producers monitor their own impacts, flexibly meet environmental targets by choosing from multiple practices, and communicate their impacts to consumers.
Article
Full-text available
The impact of dietary meat intake reduction on haematological parameters in healthy adults - Volume 76 Issue OCE3 - M. Ali, E.J. Simpson, M. Clark, A. Razak, A. Salter
Article
Background: Overconsumption of food, alcohol, and tobacco products increases the risk of non-communicable diseases. Interventions to change characteristics of physical micro-environments where people may select or consume these products - including shops, restaurants, workplaces, and schools - are of considerable public health policy and research interest. This review addresses two types of intervention within such environments: altering the availability (the range and/or amount of options) of these products, or their proximity (the distance at which they are positioned) to potential consumers. Objectives: 1. To assess the impact on selection and consumption of altering the availability or proximity of (a) food (including non-alcoholic beverages), (b) alcohol, and (c) tobacco products.2. To assess the extent to which the impact of these interventions is modified by characteristics of: i. studies, ii. interventions, and iii. Participants: Search methods: We searched CENTRAL, MEDLINE, Embase, PsycINFO, and seven other published or grey literature databases, as well as trial registries and key websites, up to 23 July 2018, followed by citation searches. Selection criteria: We included randomised controlled trials with between-participants (parallel group) or within-participants (cross-over) designs. Eligible studies compared effects of exposure to at least two different levels of availability of a product or its proximity, and included a measure of selection or consumption of the manipulated product. Data collection and analysis: We used a novel semi-automated screening workflow and applied standard Cochrane methods to select eligible studies, collect data, and assess risk of bias. In separate analyses for availability interventions and proximity interventions, we combined results using random-effects meta-analysis and meta-regression models to estimate summary effect sizes (as standardised mean differences (SMDs)) and to investigate associations between summary effect sizes and selected study, intervention, or participant characteristics. We rated the certainty of evidence for each outcome using GRADE. Main results: We included 24 studies, with the majority (20/24) giving concerns about risk of bias. All of the included studies investigated food products; none investigated alcohol or tobacco. The majority were conducted in laboratory settings (14/24), with adult participants (17/24), and used between-participants designs (19/24). All studies were conducted in high-income countries, predominantly in the USA (14/24).Six studies investigated availability interventions, of which two changed the absolute number of different options available, and four altered the relative proportion of less-healthy (to healthier) options. Most studies (4/6) manipulated snack foods or drinks. For selection outcomes, meta-analysis of three comparisons from three studies (n = 154) found that exposure to fewer options resulted in a large reduction in selection of the targeted food(s): SMD -1.13 (95% confidence interval (CI) -1.90 to -0.37) (low certainty evidence). For consumption outcomes, meta-analysis of three comparisons from two studies (n = 150) found that exposure to fewer options resulted in a moderate reduction in consumption of those foods, but with considerable uncertainty: SMD -0.55 (95% CI -1.27 to 0.18) (low certainty evidence).Eighteen studies investigated proximity interventions. Most (14/18) changed the distance at which a snack food or drink was placed from the participants, whilst four studies changed the order of meal components encountered along a line. For selection outcomes, only one study with one comparison (n = 41) was identified, which found that food placed farther away resulted in a moderate reduction in its selection: SMD -0.65 (95% CI -1.29 to -0.01) (very low certainty evidence). For consumption outcomes, meta-analysis of 15 comparisons from 12 studies (n = 1098) found that exposure to food placed farther away resulted in a moderate reduction in its consumption: SMD -0.60 (95% CI -0.84 to -0.36) (low certainty evidence). Meta-regression analyses indicated that this effect was greater: the farther away the product was placed; when only the targeted product(s) was available; when participants were of low deprivation status; and when the study was at high risk of bias. Authors' conclusions: The current evidence suggests that changing the number of available food options or altering the positioning of foods could contribute to meaningful changes in behaviour, justifying policy actions to promote such changes within food environments. However, the certainty of this evidence as assessed by GRADE is low or very low. To enable more certain and generalisable conclusions about these potentially important effects, further research is warranted in real-world settings, intervening across a wider range of foods - as well as alcohol and tobacco products - and over sustained time periods.
Article
Previous research has shown that restaurant menu design can influence food choices. However, it remains unknown whether such contextual effects on food selection are dependent on people's past behavior. In the present study, we focused on vegetarian food choices, given their important implications for the environment, and investigated whether the influence of different restaurant menus on the likelihood of selecting a vegetarian dish is moderated by the number of days on which people reported eating only vegetarian food during the previous week. In an online scenario, participants were randomly assigned to four different restaurant menu conditions-control (all dishes presented in the same manner), recommendation (vegetarian dish presented as chef's recommendation), descriptive (more appealing description of vegetarian dish), and vegetarian (vegetarian dishes placed in a separate section)-and ordered a dish for dinner. The results showed that the recommendation and descriptive menus increased the likelihood of vegetarian dish choices for infrequent eaters of vegetarian foods, whereas these effects tended to reverse for those who ate vegetarian meals more often. The vegetarian menu had no impact on the infrequent vegetarian eaters' choice but backfired for the frequent vegetarian eaters and made them less likely to order a vegetarian dish. These findings indicate that people's past behavior is an important determinant of the impact of nudging on food choices, and that achieving sustainable eating may require more personalized interventions.