Higher order risk preferences for money and food

The intervention consists of exposing participants to decision-making tasks where they make choices between binary lotteries offering either monetary or food rewards. Participants are randomly assigned to one of two conditions based on the type of food rewards:

Healthy Food Condition: Participants make lottery choices with potential rewards in units of cheese.

Unhealthy Food Condition: Participants make lottery choices with potential rewards in units of chocolate.

All participants will also participate in monetary reward tasks. The risk apportionment tasks of Deck et al. (2023) will be used.

These interventions are designed to compare risk preferences across healthy and unhealthy food items, and between food and money as reward categories.

Deck, C., Huang, R.J., Tzeng, L.Y. and Zhao, L. (2023) A Simple Approach for Measuring Higher-Order Risk Attitudes.

2024-10-09

2024-11-20

Our key outcome variable is switching points in a series of binary lottery choice tasks for money and food.

Intensity measures will be constructed similar to Deck et al. (2023)

The experimental design consists of two within-subjects tasks. In each task, participants are asked to make decisions between two options presented as binary lotteries. These lotteries are designed to evaluate participants' risk aversion, prudence and temperance under two distinct conditions: for monetary rewards and for food rewards. Each participant will complete a series of binary lotteries for both types of rewards, but the order of the tasks (money or food first) will be randomly assigned for each participant to ensure there is no systematic bias based on task sequence.

Participants are randomly assigned to one of two treatment groups:

Group 1: This group will be presented with a series of binary lotteries in which the rewards are paid in healthy food items (specifically, units of cheese).
Group 2: This group will face a similar series of binary lotteries, but their rewards will be in the form of unhealthy food items (specifically, units of chocolates).

Moreover, all subjects will make decisions over monetary lottery choice tasks, thus providing a direct comparison on how they choose when it comes to food and money.

Lottery choice tasks are based on the risk apportionment tasks of Deck et al. (2023).

At the end of the study, a randomized payment procedure will be used. The computer will randomly select one of the choices made by the participant from either the monetary or food task to determine their bonus payment. This ensures that participants approach each decision with the possibility that it may affect their final earnings. The potential bonus will range from 0 to 26 units, where each unit represents either euros (for the monetary task) or units of food (for the food task).

In addition to any bonus they may earn, all participants will receive a flat rate participation fee of 5 euros, ensuring that every participant is compensated for their time, regardless of the lottery outcomes.

Randomization done by computer (using python "random" module) at the time when participant enters the experimental software.

Randomization at the individual level.

No

0

Each subject will make 27 choices for money and 27 choices for food. Therefore, the total N of observations will be 200*54 = 10800

100 subjects in the Money + Unhealthy food treatment
100 subjects in the Money + Healthy food treatment

Based on the formula by Kupper & Hafner, 1989; Diggle et al., 2002: 30; Liu & Wu, 2005: n = 2*(z_{1-alpha/2}+z_{1-beta})^2 * (1+(M-1)*rho) / M*(d)^2, and solving for d (where d = (mu0-mu1)/sigma), with z_{1-alpha/2}=1.96, z_{1-beta}=0.8416, M=3. Based on the standard deviations for the average switch point in Deck et al. (2023) A Simple Approach for Measuring Higher-Order Risk Attitudes, we have: sigma_ra=1.52, sigma_pr=2.10, sigma_temper=1.94, and assuming rho=(0(0.3)0.9) we get: d_ra = [0.34, 0.57] of a switching point d_pr = [0.43, 0.73] of a switching point d_temper = [0.48, 0.80] of a switching point