Nudging the demand for organic food

We deliver two pieces of information to consumers. First, we give them messages about the consequences of eating conventional (non-organic) foods on health or the environment. Then, we use a color scale (traffic lights) to show how far a particular food is from being organic. Literature finds that concern about the environment and personal well-being (health) is correlated to organic food consumption (Holloway et al., 2007; Shafie & Rennie, 2012; Hidalgo-Baz et al., 2017). We intend to create the messages in a framing that facilitates behavioral change, appealing to their loss aversion to change their default choice, i.e., conventional foods, to a more beneficial one, i.e., organic foods (Díaz & Del Valle, 2016).

The second part of the intervention is the color scale, which is similar to traffic lights. The purpose of this scale is to guide consumers towards a more automatic and implicit decision-making process, reducing their cognitive load (Vlaemick et al 2014; Muller et al 2019; Crossetto et al 2019). The scale grades the intensity of production, meaning the excessive use of agrochemicals and machinery. The lower the score, the closer the food is to organic production. We assign the red color to the least organic food and the green color to the most organic one.

We will implement the intervention as a framed field experiment conducted inside farmers' markets in a store set up exclusively for this purpose. Participants receive a voucher of a fixed monetary value that they can use to select conventional fresh food. A facilitator accompanies the participants and shows them a video with the intervention's messages. Then, we give them the option of exchanging their initial food choices for more organic ones. The scale is going to be represented by colored ribbons attached to the foods' packages. In the end, the facilitator interviews participants to collect demographic and some additional information.

2021-02-22

2021-04-03

The score of participant's food basket according to a scale that grades the intensity of production.

We build an index that measures the intensity of production using an expert survey. (See details in Docs and Materials). The index scores how frequently supplementary means of production are used to increase the output per unit of the crop area. E.g., agrochemicals, mechanization of production. Seven criteria characterize food production. An expert grades each criterion from 1 to 10, where the assigned value represents the probability that a food was produced more intensive. The criteria comprise two dimensions: technological-productive and environmental. The average of all the criteria gives the value of the index.

The score of the basket is given by the weighted average of each food's index value.

We will implement a framed field experiment with two stages and five between-subjects treatments. The purpose of the experiment is to identify which messages are the most effective in encouraging the substitution of conventional (non-organic) foods for more organic foods. The target population is consumers from farmers' markets of the city of Cochabamba (Bolivia). The experiment intends to create a controlled environment that preserves some central characteristics of a real market. It takes place inside a farmers' market in a store designed exclusively for this purpose. We offer food that is usually available in farmers' markets. Each food has a conventional-organic pair. There is not going to be a price gap between conventional and organic food. However, two different foods can have different prices set at the market level. Subjects will be able to see and touch the food before selecting it.

The store has two counters. In the first stage, subjects receive a voucher and approach the first counter to select an initial food basket. Then, they receive a random treatment delivered through a video. In the second stage, subjects go to the second voucher and are given the option to exchange their food choices for more organic ones. Making a sequential choice, i.e., first of conventional food and then of organic, allows us to control for the food consumption pattern of the subjects before treatment (Muller et al., 2019).

The treatments are:
(1) (Control group) Only physical appearance and origin: After choosing the conventional food basket, subjects will see a short video that states foods from the store have different origins and appearances. Then, we will present the second counter without telling them which foods are conventional and organic.
(2) Color scale: After choosing the conventional food basket, subjects will see a short video about the production of organic, and conventional food. In the end, we will explain the color scale that grades the intensity of production, meaning the excessive use of agrochemicals and machinery. The lower the score, the closer the food is to organic production. The scale is going to be represented by colored ribbons attached to the foods' packages.
(3) Color scale and health information: After choosing the conventional food basket, subjects will see a short video about the negative consequences of consuming conventional food on health. In the end, we will explain the color scale as in treatment (2).
(4) Color scale and environment information: After choosing the conventional food basket, subjects will see a short video about the negative consequences of consuming conventional food on the environment. In the end, we will explain the color scale as in treatment (2).
(5) Color scale, health, and environment information: After choosing the conventional food basket, subjects will see a short video about the negative consequences of consuming conventional food on health and the environment. In the end, we will explain the color scale as in treatment (2).

We will use block randomization. Each session will become a block or stratum. Sessions will happen in three farmers' markets on the days they usually open. We intend to define an equal number of subjects per treatment in a session. Randomization will occur during the session through an urn that will contain the vouchers. Each one will have a bar code with the treatment number. At the beginning of the experiment, subjects will select a voucher from the urn at random, thus selecting their treatment.

Consumer (individual)

No

280 consumers of farmers' markets

280 consumers of farmers' markets

56 consumers per treatment

We performed the power analysis using simulations. The experiment can detect the change of one element of the initial food basket after treatment with a 95% power and a 5% significance level. A change of this magnitude is equivalent to a drop of 0.2 s.d. in the food basket score (Mean score: 3. Standard Deviation: 1.15)