|
Field
Experimental Design (Public)
|
Before
We will work with a meal kit company to implement a large-scale online experiment with over 100,000 of their customers in the United States. In this experiment, all meals are classified into three tiers based on the carbon footprints associated with producing the meals. Customers are randomly assigned to one of four groups: (1) a treatment group that sees menus where climate impact is represented by an abstract symbol, (2) a treatment group that sees menus where climate impact is represented by letter grades, (3) a treatment group that sees menus where the most carbon-efficient meals are labelled as "climate superstars", and (4) a control group that will see the regular menus without any climate labels. The experimental menus will be applied to 10 weeks of meals.
In addition to this large-scale field experiment with our partner company, we implement a survey experiment with participants recruited via Facebook ads. In this survey experiment, we randomly assign participants to two groups: a treatment group that sees menus where climate impact is represented by an abstract symbol, and a control group that sees menus without any climate impact labels. The meals are taken from one week of the meal kit company's menu, and participants will have a chance to win a free box with the meals they chose in the survey experiment.
|
After
We will work with a meal kit company to implement a large-scale online experiment with over 100,000 of their customers in the United States. In this experiment, all meals are classified into three tiers based on the carbon footprints associated with producing the meals. Customers are randomly assigned to one of four groups: (1) a treatment group that sees menus where climate impact is represented by an abstract symbol, (2) a treatment group that sees menus where climate impact is represented by letter grades, (3) a treatment group that sees menus where the most carbon-efficient meals are labelled as "climate superstars", and (4) a control group that will see the regular menus without any climate labels. The experimental menus will be applied to 10 weeks of meals.
In addition to this large-scale field experiment with our partner company, we implement a survey experiment with participants recruited via Facebook ads. In this survey experiment, we randomly assign participants to two groups: a treatment group that sees menus where climate impact is represented by an abstract symbol, and a control group that sees menus without any climate impact labels. The meals are taken from one week of the meal kit company's menu, and participants will have a chance to win a free box with the meals they chose in the survey experiment. Due to spending limits on Facebook recruitment and the timing of our partner company's meal cycles, we will run this off-platform experiment in two waves, each for a separate week of our partner company's meals.
|
|
Field
Planned Number of Observations
|
Before
Mealkit experiment: 133,000 customers
Survey experiment via Facebook: 6,000 participants
|
After
Mealkit experiment: 133,000 customers
Survey experiment via Facebook: 6,000 participants, to be split over an initial wave with N = 2470 and a second wave with approximately N = 3500. We run this survey experiment in two waves due to the combination of spending limits on Facebook recruitment and the timing of our partner company's meal cycles.
|