Project

The negative impact of vegetarian and vegan labels: Results from randomized controlled experiments

Alex Berke

Overview

  • We found that vegan and vegetarian labels commonly found on menu items have a significantly negative impact on consumer's likelihood to choose these menu items.
  • These findings are from randomized controlled experiments that we conducted within our community and then expanded to a larger sample of US participants from surrounding states.
  • We also did not find removing these labels on menus would negatively impact vegans/vegetarians by leading them to accidentally choose options with meat.
  • Our experiment suggests these labels should be removed from menus to normalize and encourage vegetarian and vegan eating, which is more environmentally sustainable  - overall these labels do more harm than good.

Motivation

Food systems have an important impact on environmental resources and are globally responsible for a third of greenhouse gas (GHG) emissions. Reducing consumption of animal products is a critically important challenge in efforts to mitigate the climate crisis. Despite this, meals containing animal products are often presented as the default versus more environmentally sustainable vegetarian or vegan options. 

What we did

We tested whether vegetarian and vegan labels on menu items negatively impact the likelihood of US consumers choosing these items by using a between-subjects experimental design, where participants chose a preference between two items.  Menu items were presented with titles and descriptions typical at restaurants, and a random group saw "vegan" or "vegetarian" labels in the titles of one of the two items. 

We conducted field studies at the MIT Media Lab, where people selected what to eat via event registration forms. The methodology was extended to an online study, where US consumers selected what to hypothetically eat in a series of choice questions. 

Results

Overall, results showed the menu items were significantly less likely to be chosen when they were labeled, with much larger effects in the field studies, where choice was not hypothetical. 

In addition, the online study showed male participants had a significantly higher preference for options containing meat versus other participants. Results did not indicate the impact of labels differed by gender. Furthermore, this study did not find that vegetarians and vegans were more likely to choose items with meat when the labels were removed, indicating that removing labels did not negatively impact them. 

The results suggest removing vegetarian and vegan labels from menus could help guide US consumers towards reduced consumption of animal products.

This experiment was approved by the Massachusetts Institute of Technology institutional review board (IRB): protocol 2203000615.

Overview of studies

The studies test the hypothesis that labeling menu items as vegan or vegetarian has a negative impact on the likelihood of consumers choosing these items, when compared to the menu items presented without these labels. 

This hypothesis is first tested in 2 field studies at the Media Lab. The field studies establish ecological validity because participants choose what they truly intend to eat.  The research methods are then extended to an online study where the sample pool is from the surrounding area.  This allows studying a broader population and a broader scope of menu items, and collecting demographic covariates.

Each study employs a between-subjects experimental design with two experimental conditions, "labeled" and "unlabeled", to which participants are randomly assigned.  In each study, participants select what they prefer to eat, choosing between two menu items.  The menu items are the same for the "labeled" and "unlabeled" conditions. However, the "labeled" condition shows a "(vegan)" or "(vegetarian)" label on one of the two options, while the "unlabeled" condition does not. The studies test the impact of these labels.

In the field studies, the choice is between a vegan and vegetarian option. Meat options were not included in order to align with the environmental sustainability efforts of this project. In the online study, participants make a series of choices between two options, with choices that do include meat.

Field studies

Study design

Note that participants were unaware that they were taking part in the experiment in order to avoid biased results.

The field studies were conducted through event registration forms for events hosted at the MIT Media Lab during spring and summer of 2022.  The events provided attendees with (free) food. Event organizers sent email notifications for their events to the entire Media Lab community. The emails contained links to the event registration forms, which recipients were encouraged to use to confirm their attendance. The registration forms asked participants to select their meal preference for the event.

For each event, the registration forms had 2 menu options: Option 1 was vegan and option 2 was vegetarian.  However, this was not explicitly pointed out. For each option, a title for the menu option was provided, followed by a description. These titles and descriptions were from the catering menu from which the meals were then ordered. The "labeled" and "unlabeled" versions of the forms each showed the same exact options, but in the "labeled" version a "(vegan)" label was present in the title of the vegan option. No label was present for the vegetarian option.  Figure 1 provides an example, showing the menu choice question for both the "labeled" and "unlabeled" versions of the registration form for one event.

Results

In each field study, the (vegan) option 1 was preferred by participants who saw the unlabeled form. In contrast, participants who saw the labeled form preferred option 2 instead. A chi-squared test showed this negative impact of the vegan label was statistically significant  in both field studies (p < 0.05).

Online study

Study design

This online experiment asks 5 menu choice questions, which we call choice questions 1 - 5. Each choice question asks participants to select their preference between 2 menu options, referred to as option 1 and option 2. All choice questions are between sandwiches or wraps. In order to represent a sample of choice questions over a broad range of possibilities, the choice questions differ in the types of options compared (i.e. vegetarian vs vegan vs has meat).  They are summarized by their option types as follows: 

Choice question 1: (option 1) vegetarian vs (option 2) has meat

Choice question 2: (option 1) vegan vs (option 2) has meat

Choice question 3: (option 1) vegetarian vs (option 2) vegetarian

Choice question 4: (option 1) vegan vs (option 2) vegan

Choice question 5: (option 1) vegan vs (option 2) vegetarian

This study has a between-subjects experimental design, where participants are randomly assigned to one of two conditions: "labeled" vs "unlabeled".   All participants see the same menu choice questions, but in the labeled condition a "(vegan)" or "(vegetarian)" label is shown in the title of option 1. No such labels are ever shown for the unlabeled condition, or for option 2.  An example of a menu choice question presented in the survey (choice question 5) is shown in Figure 3, with the version for the labeled condition on the left, and the version for the unlabeled condition on the right.

Results

Results are shown in Figure 4. We show results for choice question 5 separately, as it most directly extends the field studies by asking participants to choose between a vegan versus vegetarian option.

Similar to the field studies, labels had a negative impact on choices, however the effects are smaller. Again, a chi-square test was used to show the negative impact of labels was statistically significant (p < 0.05).


More information about our study, methods, and results can be found in our paper published in the journal Appetitehttps://doi.org/10.1016/j.appet.2023.106767

This experiment was approved by the Massachusetts Institute of Technology institutional review board (IRB): protocol 2203000615.