Methodology

1 Study identification

1.1 Scope

LIME contains evidence from intervention studies that aimed to reduce animal product consumption. We included results from studies that matched the following criteria.

Publication types: Peer-reviewed scientific publications, but also unpublished preprints, theses, or reports by advocacy groups and other organizations.

• We include publications in languages we are fluent in (English, Dutch, German).

Samples: Most types of populations and sampling procedures.

• We include studies with participants from any country and of any age, who were recruited online or in-person, via market research companies or panel providers (e.g., Prolific, Kantar), university student participant pools, or by directly approaching or observing people in supermarkets, restaurants, or university dining halls.
• We exclude studies with specific participant inclusion criteria that can be expected to artificially deflate or inflate the observed effectiveness of an intervention. For example, we exclude studies where an intervention was tested only among people who accepted a flyer on animal cruelty, who pledged to reduce their meat consumption in the following month, or who indicated in an earlier survey that they are very open to reducing their meat consumption. This does not apply to studies that were conducted in locations with people who are more open to reducing their animal product consumption (e.g., we include studies with students from liberal arts colleges).
• We exclude studies that recruited patient samples (i.e., only participants with a particular mental or physical illness).

Study designs: Experimental and quasi-experimental studies (e.g., pre-post designs) with a neutral control condition.

• We exclude studies that compare two interventions without a separate control condition (e.g., health message vs. environmental message).
• In lab studies, participants in control conditions were usually exposed to a neutral control stimulus (e.g., reading a text on an irrelevant topic) or to no stimulus at all. In some study settings, a truly neutral or passive control is not feasible or possible. For example, a menu in a restaurant has to list some item first. These conditions are treated as eligible control conditions if they arguably resemble the current default (e.g., the first entrée on a menu is typically a meat option, there are typically more meat than non-meat options on a restaurant menu).

Intervention types: Any intervention that aimed to (or can be expected to) reduce animal product consumption.

• We include more generic interventions that do not directly target animal product consumption if they could plausibly reduce overall animal product consumption (e.g., promoting a Mediterranean diet). We exclude them if this is not the case (e.g., promoting a sustainable lifestyle in general would be excluded because people do not necessarily know that this entails lower animal product consumption).
• We exclude interventions that aim to shift consumption from some animal products to others (e.g., promoting a shift from beef to chicken, fish, or insects).
• We exclude interventions that aim to increase animal product consumption.
• We include interventions that change the name or label of products or dishes, but only if two conditions are met. First, there is a dish name that can be considered a neutral control. "Vegetarian lasagna" or "veggie burger" are common enough to serve as valid controls. Other labels, like "garden pasta" (not common, might evoke positive emotions) or "leek quiche" (not obvious that it is vegetarian) would not be valid controls. Second, the alternative dish name can be reasonably expected to increase selection. It is not clear if "meat-free" or "plant-based" would increase selection, so these would not be valid interventions. Other labels, like "fresh garden pasta" or "juicy flame-grilled burger" would count as valid interventions because they can be expected to evoke positive associations or emotions.
• Sometimes changes in consumption are tested in response to events that were not experimentally manipulated by the researchers. We include these studies if the event could plausibly be caused intentionally as an intervention strategy (e.g., news coverage of poor animal welfare conditions on farms). We exclude them if this is not the case (e.g., an E. coli contamination of meat products).
• We include studies with interventions that target multiple pathways (e.g., information on the negative consequences of meat consumption on animal welfare, the environment, and health).

Outcome measures: All measures broadly related to animal product consumption.

• We include data on observed behaviors (e.g., sales data, observed meal choices), self-reported behaviors (e.g., food diaries, self-reported diet), intentions (e.g., reported intentions to reduce future meat consumption), and attitudes or beliefs (e.g., attitudes on or moral judgments of meat consumption).
• We exclude more general attitudes and intentions related to animal welfare (e.g., endorsements of animal rights), health (e.g., intentions to eat a healthier diet), or the environment (e.g., intentions to reduce one's greenhouse gas emissions).
• We exclude outcomes that combine plant-based and animal product choices (e.g., a sustainability index where beans but also fish result in a higher score).
• We only include measures that focus solely on vegetarian products (e.g., purchases of meat replacements) if this is measured in a context where choosing it can be expected to replace animal products. For example, we exclude studies that measure liking of plant-based steak or intentions to eat more beans because this does not necessarily imply that people will reduce their consumption of animal products.

1.2 Search strategy

The typical approach to identifying relevant studies for a meta-analysis involves (a) specifying a string of search terms that match what is contained in the title and abstract of relevant studies, (b) entering the search terms into one or multiple databases of academic articles, and (c) screening all papers that are flagged by this search. This approach was not feasible for the present project. To create a comprehensive overview of the literature, we include various intervention types, outcomes, and study designs. Studies that we intend to include are conducted in psychology, economics, consumer behavior, climate science, nutrition science, and other fields. Search terms that capture all relevant studies across diverse fields, topics, and methodologies have to be so broad that they would also capture many irrelevant studies on adjacent topics (e.g., thousands of studies on changing the taste or other attributes of animal products, or on the effect of diet on health outcomes).

Similar to a previous meta-analysis on this topic (Green et al., 2026), our study identification strategy leveraged the availability of many systematic (narrative) reviews on various topics related to meat reduction interventions. In short, our search strategy primarily relied on (a) identifying relevant reviews and (b) extracting papers cited within these reviews. We describe our strategy below. Note that new research on this topic is published every week or so and our goal is to keep updating the database.

For our primary search strategy, we identified 2 relevant meta-reviews (a narrative review of reviews). These meta-reviews cited 41 reviews, of which 25 reviewed evidence on interventions aimed at reducing animal product consumption. 16 reviews were excluded because they focused on interventions to increase calcium intake (Jung et al., 2016), on the acceptance of alternative proteins (which does not necessarily entail a reduction of animal product consumption; Siddiqui et al., 2022), or on other irrelevant topics. From each included review, we extracted all papers that were reviewed, leading to a total of 1,757 identified papers.

As a secondary search strategy, we used various channels to identify additional reviews and papers. These papers came to our attention via announcements on social media or suggestions by other researchers. After curating an initial list of papers that were deemed relevant or irrelevant, we also used a large language model to identify additional papers not included in the list. This search turned up 7 additional reviews and collections of studies (1 additional review was excluded), from which we extracted 531 papers. We also identified 132 additional papers. Overall, our secondary search strategy identified 663 papers.

Combining papers from both search strategies led to a list of 2,420 papers. After removing 1,319 duplicate entries and 17 papers which could not be found or accessed, we were left with 1,084 papers to be screened for relevance.

1.3 Screening

Of these papers, 1,050 have been screened so far, considering the inclusion criteria outlined in section 1.1, based on the abstract and, in case the information in the abstract was insufficient, based on the description of the study's research question, hypotheses, and methods. The remaining 34 papers have yet to be screened. A total of 876 papers were excluded: 806 papers because they did not meet the inclusion criteria and 70 papers because, even though they were deemed relevant, the statistics that are necessary for our analyses were not reported in the paper (e.g., means, sample sizes, standardized effect sizes). We are in the process of contacting the authors of these papers to obtain the relevant statistics and include the studies in our database.

This screening process leaves us with the studies currently in the database. These numbers are updated automatically as the database grows.

Many studies included multiple estimates of intervention effectiveness, for example, because multiple interventions or outcomes were tested.

2 Study coding

Each study included in the database was coded on dozens of dimensions. We extracted information at the

• Paper level: publication type, year, journal, accessibility (paywalled vs. open access), etc.
• Study level: total sample size, data availability, preregistration availability, study design, randomization procedure, etc.
• Sample level: sample size per condition, the country in which the study was run, the recruitment location (e.g., university, online panels), etc.
• Condition level: intervention type, intervention duration, intervention medium, etc.
• Outcome level: outcome type (e.g., intentions, behavior), outcome subtype (e.g., meat consumption, self-reported diet), measurement type (e.g., food diary, sales data), days since intervention.

The codebook, which you can download from the Meta-analysis page, provides an overview of all variables. Most variables and attributes are self-explanatory. Below, we describe three coding approaches in more detail because they are not self-explanatory and are central to the analyses.

2.1 Coding study designs

The studies that are included in the database vary substantially in their design (e.g., whether results for the control and intervention conditions come from the same or different groups of participants). We capture this variability with eight variables.

We distinguish between two ways of condition assignment.

• Individual: Each participant is assigned to a condition. This is the typical procedure in lab or online studies, where each individual is assigned to a condition (or in a within-subject design, to multiple conditions or multiple measurement points for the same condition).
• Cluster: In some studies, groups of participants are assigned to different conditions. For example, all participants who show up for a lab study or accept a leaflet on a specific day might be exposed to the same treatment. Researchers might also assign conditions to restaurants, school classes, or other clusters.

Cluster label describes to which type of cluster conditions were assigned (e.g., restaurant, day).

Cluster number shows the total number of clusters to which conditions were assigned (e.g., how many restaurants were assigned in total).

We distinguish between three condition assignment processes.

• Random: Participants are randomly assigned to conditions.
• Non-random researcher-assigned: Researchers assigned participants to conditions in a way that was not fully random and that might introduce systematic differences between participants across conditions (e.g., control leaflets are distributed in week 1, intervention leaflets in week 10).
• Self-selection: Participants' presence in the conditions is (partly) resulting from their own decisions (e.g., participants signed up for university classes during which the climate impact of meat is or is not discussed, participants' data is only recorded for a study if they had previously accepted a leaflet).

We distinguish between four types of condition assignment structures.

• Between: Different groups of participants are assigned to the different conditions.
• Within: The data comes from the same group of participants (e.g., a pre-post design with data from before and after an intervention).
• Crossover: A special case of a within-subject design in which participants provide data for multiple conditions which are experienced at different time points.
• Mixed: The study design combines a between component (e.g., different participants in intervention and control conditions) and a within component (e.g., participants in all conditions provide data before and after being exposed to the intervention).

Data tracking describes whether it is possible to match a specific data point to a specific participant. This is usually the case in a typical lab study in which individual participants are assigned to one study and their data can be matched to an identification number created by the researcher or the survey program. It is not the case in studies that analyze, for example, the sales data of a restaurant or supermarket. In these data sets it is not possible to tell (a) if multiple observations come from the same person (e.g., many meat items sold by a supermarket were bought by the same person) and (b) if the same person was even exposed to multiple study conditions (e.g., the same person visited the control and the intervention supermarket). Both issues, sometimes referred to as pseudoreplication and cross-contamination, complicate interpretations of the results.

Finally, study RCT indicates whether a study shows a specific set of design features that are associated with high-quality randomized controlled trials. A study gets this label if three conditions are met:

• The condition assignment process is random.
• The study has a between or mixed condition structure.
• Data tracking is possible.

2.2 Coding intervention types

A primary goal of this project is to estimate the effectiveness of different types of interventions. We categorized interventions according to the pathway via which they are expected to change animal product consumption. Given the diversity of interventions that have been tested and the diversity of terminology in the literature, there is no obviously correct way to categorize interventions. Our coding scheme was based on three considerations.

First, our intervention categories primarily reflect the specific pathways (most could be described as psychological mechanisms) via which the intervention is expected to change behavior. We believe that a focus on psychological mechanisms is more fruitful for understanding "what works". We code what is being done (e.g., distribution of flyers, classroom discussions, PowerPoint presentations), but we focus more on how these activities are expected to change behavior (e.g., information on cruel practices in factory farms, on social norms, or on concrete strategies to incorporate more plant-based products into one's diet).

Second, the breadth of our categories is informed by the availability of studies. It would be informative to know the effectiveness for very specific intervention categories (e.g., providing health information via a product label, a weekly text message, etc.), but there currently aren't enough studies so that we can derive precise effectiveness estimates for such narrow categories. Very broad categories (e.g., all information-based interventions) would mean that we have many studies per category. However, such broad categories would conflate very different pathways that likely differ in their effectiveness (e.g., providing health vs. animal welfare information). Our coding strategy attempts to balance these concerns. Our primary intervention categories are broad enough so that effectiveness estimates for almost all categories are based on dozens of studies and not just a few. They are narrow enough so that they do not conflate very different psychological pathways via which the intervention is expected to change behavior. Some intervention types were deemed psychologically similar enough to be combined in one category (e.g., increasing the availability or variety of vegetarian options), and this is also reflected in how they are labeled (e.g., "availability/variety/size").

Third, this means that in some cases, our labeling of an intervention deviates from how the intervention was labeled by the researchers who conducted the study.

Below are descriptions and examples for the 23 categories into which intervention studies are currently organized. These intervention categories are organized into six broader groups. This organization might change in the future, if we deem it informative to split up certain categories due to the availability of more evidence. Note that categories are not exclusive. Many interventions target multiple mechanisms (we refer to these as multi-component interventions), and this is also reflected in our coding. Many categories contain interventions that either steer people away from meat consumption, or toward vegetarian alternatives, or both simultaneously. However, when describing intervention types below, we mostly focus on the anti-meat framing for brevity.

2.2.1 Persuasion interventions

Interventions that primarily aim to change behavior by presenting relevant factual information.

Health information

Interventions that provide information on the negative effects of meat consumption on health outcomes.

• A chatbot sent different daily persuasive messages on the health benefits of eating little red and processed meat to participants (e.g., "If you eat little red and processed meat, you will protect your health from colon cancer/heart disease/respiratory disease"; Carfora et al., 2019).
• Participants saw a meat option with a pictorial and textual warning label that eating meat contributes to poor health. The text contained references to the scientific sources where the warning messages pertaining to health consequences of meat consumption came from (Hughes et al., 2023).

Environmental information

Interventions that provide information on the negative effects of meat consumption on the environment and climate change.

• Participants saw a screen with a picture of a veggie burger and the label "Sustainable Choice." The text on this screen read, "Many people don't know how much our food choices impact the environment. Eating sustainable foods is a good way we can fight climate change! The Garden Fresh Veggie Burger is an especially sustainable choice. Consider giving the Garden Fresh Veggie Burger a try today!" (Piester et al., 2020).
• Participants were shown a 25-minute sequence out of the documentary "Cowspiracy" in which various environmental impacts of meat production were presented, such as greenhouse gas emissions, water consumption, or water pollution (Bschaden et al., 2020).

Animal welfare information

Interventions that provide information on the negative effects of meat consumption on animal welfare.

• Participants were given a newspaper article with fictional results of a scientific investigation, reporting the negative impact of meat consumption on animal welfare (Cordts et al., 2014).
• Participants read a message concerning poor animal welfare in intensive farming (e.g., "A lack of respect for their emotional state exacerbates the situation. There is no shortage of incidents of physical and sexual violence, leading to constant experiences of fear."; Carfora et al., 2025).

Taste/disgust information

Interventions that provide information on the tastiness of vegetarian food, or on aspects of meat consumption that are meant to evoke disgust.

• Participants read a persuasive essay describing disgust-evoking aspects of meat production and consumption (e.g., "...ABC News recently broke a story highlighting the amount of 'pink slime' that is added to the ground beef that you see in grocery stores, restaurants, and even school cafeterias..."; Palomo-Vélez et al., 2018).
• After buying their noodles, consumers could add fish sauce from a bottle. An A4-sized information sheet with an affect-provoking picture of a girl making a disgusted face with a short message saying "too salty" was displayed nearby the fish sauce bottle (Kanchanachitra et al., 2020).

2.2.2 Social influence interventions

Interventions that primarily aim to change behavior via conformity with social norms, via activating image concerns, or via similar social processes.

Authority/role models

Interventions that show authority figures or role models who advocate against meat consumption.

• Participants saw a social media post in which a masculine model was shown alongside a plant-based meal (Fonseca et al., 2025).
• Alongside other information, participants read that many celebrities started to reduce their meat consumption (Fesenfeld et al., 2023).

Descriptive norms

Interventions that provide information on the absolute prevalence of a relevant attitude or behavior (e.g., "most people eat plant-based at least once a week").

• Participants entered their weekly servings of food purchased across 13 categories in an app and received feedback on what percentage of people have more or less emissions than them (Bronnenberg et al., 2025).
• Self-service order terminals showed: "A lot of our customers have tried a Plant-Based Product. 0% meat, 100% taste." (Edwards et al., 2026).

Dynamic norms

Interventions that provide information on the increasing prevalence of a relevant attitude or behavior (e.g., more and more people are reducing their meat consumption).

• People saw the following message: "We've noticed that more and more College [X] students are choosing veggie lunches." In cafeterias where meal selections were made in the cafeteria, messages were printed on A4 posters and displayed at the point of decision-making in at least two highly visible locations (Biggs et al., 2026).
• Participants saw an online menu with a centered note at the top of the menu including a dynamic norm message: "We've noticed that customers are starting to eat less meat by choosing more meatless dishes. We're committed to ordering a menu that gives you a variety of choices, all with fresh ingredients" (Sparkman et al., 2020).

Identity/reputation

Interventions that activate image concerns, or that highlight how desirable traits or values favor vegetarian choices.

• Participants were presented with an image of a zoo or hospital accompanied by a text asking them to imagine being at the zoo or hospital, "During your visit you get hungry and decide to have lunch in the restaurant of the zoo/hospital," followed by a question of whether they consider animal welfare to be important (Bouwman et al., 2022).
• Participants were made aware of their prejudice towards animals by conducting and receiving a bogus result from an Animal-Human Implicit Association Test (IAT), suggesting they preferred humans over animals, an attitude explained as contributing to harm toward animals (Stel & Unterweger, 2025).

2.2.3 Affect and association interventions

Interventions that primarily aim to change behavior by triggering relevant emotions or associations.

Hedonic labeling

Interventions that use alternative names for dishes that are meant to promote vegetarian choices.

• Participants were presented with a menu including plant-rich dishes with appealing names, such as Western Australian Sunkissed Tomato Barley Risotto, Queensland Roasted Tomato and Charred Capsicum Soup, and English Garden Salad (Gavrieli et al., 2022).
• The name for the vegetarian breakfast was changed from "meat-free breakfast" to a more appealing name that did not emphasize the absence of meat: "field-grown breakfast" (Bacon et al., 2018).

Origin of animal product

Interventions that create a connection between an animal product and the animal that it is coming from.

• Participants were presented an image of a cooked meat dish paired with an image of a calf. The animal was presented above the meat dish on the page and participants were told the meat "comes from the baby animal depicted above" (Piazza et al., 2018).
• Participants saw a beef dish recipe in a short text with a photo of a cow, illustrating the source of beef in the dish (Tian et al., 2016).

Perspective taking/individuation

Interventions that introduce individual animals or that prompt participants to imagine what it's like to be an animal.

• Participants saw a picture of a pig playing with a soccer ball and were given the instruction to "imagine that you are the pig in the picture. Your favorite sport to play is soccer; you love working on your dribbling and watching games on TV. Write a brief story describing some of the other activities you love to do." (Johnson et al., 2021).
• Participants saw a series of Facebook posts showing the special characteristics of animals, which were presented as unique individuals (Bertolaso, 2015).

Priming

Interventions that expose participants to visual or other sensory stimuli that are meant to induce emotions, thoughts, or goals conducive to plant-based food choices.

• Participants viewed a slideshow consisting of 15 stock photographs portraying scenes of vulnerability, care-giving, human connection, and times of emotional distress of animals, which were shown to induce compassion in a pretest (Pohlmann, 2021).
• The food environment was tailored to facilitate the choice of plant-based options, including a green ambience of plants, green serving bowls, and herbs in the dining area. The fresh herbs provided a strong odour and participants could freely use the provided fresh herbs (basil, parsley) as part of their meal (Friis et al., 2017).

2.2.4 Choice architecture interventions

Interventions that primarily aim to change behavior by changing the structure of the situation in which food choices are made.

Default

Interventions that made the vegetarian option the default choice.

• Individuals responded to an RSVP survey for an event in which the vegetarian option was the default meal preference (Boronowsky et al., 2022).
• The bean burger was presented as the default option, but in a separate box next to it: "Rather have a beef burger? This is also possible on request." (Taufik et al., 2022).

Visibility/salience/ease

Interventions that make the vegetarian options more visible, salient, or easier to choose.

• Customers were presented with a digital menu in which the icon representing the "green" food category was moved from the sixth position in the digital menu grid (out of ten) to the first position to increase its visual salience (Reinholdsson et al., 2023).
• The vegetarian option was highlighted as the dish of the day on the menu and by the waiter or waitress (Saulais et al., 2019).

Availability/variety/size

Interventions that increase the number or diversity of vegetarian options that are offered, or that decrease the amount of meat in dishes.

• The portion sizes of vegetables on the plates of main dishes were doubled (150 g of vegetables instead of 75 g) and the portion sizes of meat were reduced by an average of 12.5% (Reinders et al., 2017).
• One meat and two vegetarian hot lunches were offered, instead of only one vegetarian meal (Biggs et al., 2026).

2.2.5 Goal pursuit interventions

Interventions that primarily aim to change behavior by helping people to form and pursue relevant goals and intentions.

Planning/pledge/reminder

Interventions that prompt people to pledge to change their meat consumption, that help people plan how they are going to make a change, or that remind people of their plans.

• Participants met with a dietitian who made specific suggestions on how to incorporate soy foods into the daily menu, such as replacing meat with tofu in stew or fried dishes, cow milk with soy milk, and cookies with a soy protein bar (Acharya et al., 2004).
• Every day for two weeks, participants received messages, which also included a reminder to try not to eat more than two portions of red and processed meat each week (Wolstenholme et al., 2020).

Efficacy/consequences/feedback

Interventions that highlight the positive consequences of reduced meat consumption or provide feedback on people's progress.

• For 30 days, participants' goal to reduce animal product consumption was supported with an app which included, among other things, a habit-tracking tool and feedback on saved animals, land, water, and emissions (Severijns et al., 2024).
• Participants received the following message before making a dish selection: "Each of us can make a positive difference for the planet. Swapping just one meat dish for a plant-based one saves greenhouse gas emissions that are equivalent to the energy used to charge your phone for two years. Your small change can make a big difference." (Blondin et al., 2022).

Food provision

Interventions that provide free meat replacements or other vegetarian foods.

• Participants received free plant-based meat alternatives for four weeks (Bianchi et al., 2022).
• Participants received a regular supply of soy products free of charge (Acharya et al., 2004).

Cooking skills

Interventions that facilitate vegetarian meal planning and cooking.

• Participants completed a veg*n cooking workshop at school, ate the dish they cooked, and joined a group discussion about the cooking process (Jans et al., 2023).
• Participants received by email a cookbook of ten easy-to-prepare and cheap meat-free recipes (Marty et al., 2025).

2.2.6 Other interventions

Interventions that did not fit into the other categories.

Taste

Interventions that aim to improve the taste of vegetarian dishes.

• A two-day workshop was conducted with the canteen kitchen, which included practical demonstrations of how to improve the visual and sensory appeal of plant-based meals (Guedes et al., 2023).
• Higher-end plant-based meats (from Impossible Foods) were offered in a cafeteria (Malan, 2020).

Price

Interventions that reduce the price of vegetarian options relative to meat options.

• The college cafeteria decreased the price of the vegetarian option by 20 pence (from £2.05 to £1.85) and increased the price of meat options by 20 pence (from £2.52 to £2.72; Garnett et al., 2021).
• Participants received a voucher worth $1.40 if they chose a meat-based dish, or $2.80 if they chose a vegetarian dish (Pizzo et al., 2026).

Other

Interventions that did not fit into one of the categories.

2.3 Coding outcome measures

We use five variables to describe the type of outcome measure that was used in a study.

We distinguish between four outcome categories.

• Attitudes/beliefs include outcomes such as scores on a meat attachment questionnaire (Bianchi et al., 2022), agreement with statements like "eating less meat, chicken and fish (in total) is the right thing to do" (Mercy for Animals, 2016), and anticipated enjoyment of plant-based dishes (Malan, 2020).
• Behaviors (observed) include outcomes such as vegetarian sales records from restaurants (Reinholdsson et al., 2023) or university cafeterias (Schwitzgebel et al., 2020), meal choices at events (Boronowsky et al., 2022), and the selection of vegetarian meal plans at university (Stewart, 2021).
• Behaviors (self-reported) include outcomes such as self-reported changes in meat consumption (Aberman & Plaks, 2022), and the frequency of animal product consumption as assessed with multi-day food diaries (Animal Charity Evaluators, 2013) or 24-hour recalls (Feltz et al., 2022).
• Intentions include outcomes such as self-reported intentions to reduce future meat consumption (Pauer et al., 2022) and hypothetical choices of vegetarian dishes (Krpan & Houtsma, 2020).

These categories are divided into 22 outcome subcategories.

Attitudes/beliefs

• Meat attitude: general evaluations of meat or meat eaters.
• Meat hedonic: perceived tastiness and anticipated enjoyment of meat products.
• Moral judgment: judgments about the ethicality of eating meat.
• Vegan hedonics: perceived tastiness and anticipated enjoyment of vegan products.
• Vegetarian attitude: general evaluations of vegetarian food or vegetarians.
• Vegetarian hedonics: perceived tastiness and anticipated enjoyment of vegetarian products.

Behaviors (observed)

• Animal product consumption: observed consumption of animal products.
• Meat consumption: observed consumption of meat.
• Vegan consumption: observed consumption of vegan food.
• Vegetarian consumption: observed consumption of vegetarian food.

Behaviors (self-reported)

• Animal product consumption: self-reported consumption of animal products.
• Dairy consumption: self-reported consumption of dairy.
• Diet: self-reported diet (e.g., omnivorous, vegetarian, vegan).
• Egg consumption: self-reported consumption of eggs.
• Meat consumption: self-reported consumption of meat.
• Vegan consumption: self-reported consumption of vegan food.
• Vegetarian consumption: self-reported consumption of vegetarian food.

Intentions

• Animal product intentions: self-reported intentions to change one's consumption of animal products.
• Diet intentions: self-reported intentions to change one's diet (e.g., omnivorous, vegetarian, vegan).
• Meat consumption intentions: self-reported intentions to change one's consumption of meat.
• Vegan consumption intentions: self-reported intentions to change one's consumption of vegan food.
• Vegetarian consumption intentions: self-reported intentions to change one's consumption of vegetarian food.

Some measures specifically target certain foods, such as the consumption of red meat or eggs. This is captured with a product type variable that shows whether a measure captured a specific category of animal products (meat, eggs, dairy), and a product subtype variable that shows whether a measure captured a more specific subcategory (beef, cheese, chicken, dairy, fish, lamb, milk, pork, poultry, processed meat, red meat, red and processed meat, seafood, shrimp, turkey, white meat, yogurt).

Finally, a measurement type variable indicates what type of measurement instrument was used to record the outcome.

• Food diary: questionnaires that allow people to record their food consumption in greater detail (e.g., 24-hour recalls of all food that was consumed, food frequency questionnaires that capture the frequency of consuming a wide range of products).
• Meal choice: real or hypothetical choices of specific dishes or products.
• Sales data: records of products sold in a restaurant, supermarket, or another venue.
• Survey: questionnaires that capture self-reported attitudes, beliefs, or intentions with single items or multi-item scales.