We are happy to announce that all trial registrations will now be issued DOIs (digital object identifiers). For more information, see here.
Encouraging low-carbon Food Consumption using Collaborative Game-Mechanisms
Last registered on November 25, 2019


Trial Information
General Information
Encouraging low-carbon Food Consumption using Collaborative Game-Mechanisms
Initial registration date
December 17, 2018
Last updated
November 25, 2019 2:44 AM EST

This section is unavailable to the public. Use the button below to request access to this information.

Request Information
Primary Investigator
University of Kassel
Other Primary Investigator(s)
PI Affiliation
University of Kassel
Additional Trial Information
On going
Start date
End date
Secondary IDs
The study comprises a field experiment using gamification in order to encourage individuals to consume food in a more climate-friendly way. The digital game-app Carbonia features collaborative social interaction mechanisms and is played among university friends. We aim to analyze how the game interacts with the structural evolution of the friendship network of the participants and the network mechanisms that influence dietary choices.

Given that the reduction of meat and dairy products is seen as an important part of climate mitigation activities necessary to achieve the 2-degree climate target (e.g. Bryngelsson et al., 2016; Girod et al., 2014), private food choices can actively contribute to climate change mitigation efforts.
External Link(s)
Registration Citation
Köbrich León, Anja and Janosch Schobin. 2019. "Encouraging low-carbon Food Consumption using Collaborative Game-Mechanisms ." AEA RCT Registry. November 25. https://doi.org/10.1257/rct.3643-3.1.
Former Citation
Köbrich León, Anja, Anja Köbrich León and Janosch Schobin. 2019. "Encouraging low-carbon Food Consumption using Collaborative Game-Mechanisms ." AEA RCT Registry. November 25. https://www.socialscienceregistry.org/trials/3643/history/57689.
Experimental Details
Given that food intake was shown to be a social action and that mitigating GHG emissions is a “collective risk social dilemma” since “[r]eaching the collective target requires individual sacrifice” (Minski et al., 2008, p. 2291), we develop a digital social interaction game: Carbonia. The game is intended to do exactly that: individual actions regarding low-carbon food consumption must be coordinated in order to achieve a common goal, namely to save Carbonia from destruction. This is only possible if the individual climate-relevant food decisions are coordinated.
Game description: Humans arrive at an unexplored planet. They start building farms and grow animals through ingredients obtained by uploading images of real meals and evaluating them. The abstract animals protect the farms in the game world. Several times a day monsters arrive to destroy the works of the people. The world of all players is connected, so they can go down together, if the monsters are successful. The players get resources by uploading pictures of their meals and rate others. These resources are used to build and improve animals that defend the players. Through these operations, conclusions on behavior can be drawn, when the game mechanism is accepted by the players. This game is played in the landscape-modus on a smartphone The success of the game is based on the interaction of the players and the uploading of food images. The game is a circular process. Meals that the players eat should be photographed and uploaded. Then the player should indicate the composition of the food by evaluating the proportions of the three main ingredients. Players can choose from 16 food categories, divided into 4 main categories (meat, cereals, vegetables and dairy products).
Summing up the social game Carbonia does not manipulate the information presented to citizens to change their attitudes and behavior without asking them for their consent, such as nudges; but, if consented to, it reorganizes the information presented to the players in a way that is beneficial for changing their attitudes and their behavior without necessarily being transparent in this effort. This makes our approach different from governance instruments such as educational campaigns that are based on the transparent presentation of relevant information to the citizens. Social games, such as Carbonia, can thus be understood as a new semi-paternalistic, highly scalable, participative governance instrument based on consented manipulation that allows designing policies to face global change.
The game intervention is further accompanied by a panel network intervention study in order to assess the mechanics and (individual) factors driving the behavioral change. We are thus able to link individual decision making and game behavior to individual characteristics as well as to changes in the social network of the individuals.
Intervention Start Date
Intervention End Date
Primary Outcomes
Primary Outcomes (end points)
Mainly three points
1. Evolving low-carbon food choices / curtailment behavior and efficiency knowledge with regard to climate relevance of food in network
2. Effect of the intervention
3. Spill-over of individual behavior
For the intervention study dependent variables are either food choices (measured by a discrete choice experiment (DCE)) or individual knowledge (assed through a self-assed knowledge test on efficiency of food choices in order to mitigate CO2e emissions.
Explanatory variables contain demographic information (age, gender) friendship (number of friends within / outside university, meat eating behavior of friends, connectedness to friends) as well as individual preferences (risk, trust, self-esteem, environmental attitudes and locus of control).
Primary Outcomes (explanation)
Secondary Outcomes
Secondary Outcomes (end points)
Secondary Outcomes (explanation)
Experimental Design
Experimental Design
• The eating behavior/food knowledge of friends influences egos behavior (social influence process).
• Similarity in eating behavior/ food knowledge makes friendship ties between alter and ego more likely by making ties more durable once they have been established (selection process by similarity endowment).
• There are unobserved time constant confounders of the social influence and the social selection process.
• There are no unobserved time varying confounders of the social influence and the social selection process.
• We assume that the causal effect between the social network (via the selection process) and the individual behavior/knowledge (via the influence process) are cross-lagged, i.e. they are not simultaneous.
• There is the possibility of two sided non-compliance, i.e. people can reject the treatment and people who were not intended to be treated can get the treatment
• The two-sided non-compliance is influenced by decisions of ego’s friends to accept or reject the treatment: the more friends are offered the treatment, the more a person will try to obtain it too and the more inclined they will be to accept it.
• The two-sided non-compliance can be affected by an unobserved time constant confounder that affects the outcome variable and the network selection process.
Experimental Design Details
Not available
Randomization Method
Observational Study:
All master students of the University of Kassel as well as bachelor students from two to three Chilean universities (yet to be determined) have the possibility to take part in the survey.
Intervention (natural field experiment as a randomized trial with non-compliance)
Participants of the survey are randomly assigned to one out of two treatment groups in equal shares. Treatment group one has the possibility to play the game. The second group gets specially designed newsletter that gives them information on the climate-relevance of food choices.
Spillover-Experiment (Framed field experiment as a randomized trial)
Using a factorial-design, participants of the survey are randomly assigned to one out of two groups. The first group have to decide how much of 50 €, which they can win with a probability of 1:20, they want to give to an environmental origination (Deutsche Umweltstiftung) and afterwards which amount of another 50 € the can win with the same probability they want to donate to a social organization (Welthungerhilfe). The second treatment group firstly decides about how much to give to the social organization and afterwards to the environmental organization.
Randomization Unit
Individual (students)
Was the treatment clustered?
Experiment Characteristics
Sample size: planned number of clusters
two clusters (University of Kassel, Chilenean University)
Sample size: planned number of observations
The population of inference for the survey consists in the master students of the university of Kassel and of the bachelor students of two or three Chilean Universities (participating universities are still to be determined). The students are offered an online and paper-and-pencil questionnaire via a mailing list of the university and within their respective major courses. We expect a quota of approx. 850 questionnaires in the first wave of the first location (Germany) and 1000 to 2500 questionnaires in the first wave of second location (Chile) (see power calculations for detail).
Sample size (or number of clusters) by treatment arms
for University of Kassel: approx 750 game intervention, 750 newsletter
for Chile pprox. 1200 game and 1200 newsletter
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
see PAP
IRB Name
German Association for Experimental Economic Research
IRB Approval Date
IRB Approval Number
Analysis Plan

There are documents in this trial unavailable to the public. Use the button below to request access to this information.

Request Information