Experimental Design
This project aims at investigating the effect of different incentives or choice architectures on individuals’ food choice in terms of diet health.
Decision Task
To examine individuals’ food choices, we adopt the experimental framework of Sadoff, Samek, and Sprenger (2020). Subjects are given lists of 20 food items presented in random order, in which 20 items are fruits and 20 items are snacks. These 20 items are presented in a table of 4 rows and 5 columns. The proportion of fruits and the nutrition index of the selected basket measure the diet health. In total, we construct three lists with 60 distinct items of food.
Control and Treatments
We design eight conditions. In the Baseline conditions, subjects make decisions on the three lists sequentially without variations. The remaining seven conditions involve treatment variations. Specifically, the first lists are identical across all eight conditions, and we start treatment variations from the second list.
In the Financial Incentives conditions, subjects are given an additional five yuan if they select more than five fruits in one list, and if they select more than five snacks in the other list. Put differently, we consider both the nudge for a healthy diet and the nudge for an unhealthy diet, which allows further discussions on the symmetry of individuals’ responses to nudges.
The provision of financial incentives is a classical method to solve the self-control problem. Choice architecture is another approach, which concerns the design of choice presentation without distorting individuals’ incentives. Our remaining treatment conditions use different methods from choice architecture.
The Information condition provides subjects with information on the number of calories of every 100g of food for each item. The Half Portion condition requires subjects to choose between half portion or full portion of the food when they pick an item. The Default condition sets default choices for subjects and subjects can adjust selection easily. In one list, we randomly pick seven fruits and three snacks as the defaults, while in the other list, we randomly pick three fruits and seven snacks as the defaults. In the Recommendation condition, we provide recommendations for subjects by labeling ten items as “recommended”. In one list, a random set of seven fruits and three snacks is recommended, while in the other list, a random set of three fruits and seven snacks is recommended. In the Bundle condition, we provide a bundle of ten items to subjects, in which we show the name of these ten items selected from the list. In one list, the bundle contains a random set of seven fruits and three snacks, while in the other list, the bundle contains a random set of three fruits and seven snacks. In the Order condition, we group all fruits together and all snacks together instead of randomizing all items. In one list, the group of fruits is presented in the first two rows, in the other list, the group of snacks is presented in the first two rows.
Incentives:
All subjects receive a participation fee. We will randomly select one out of every ten subjects to receive gifts. For the chosen subjects, we will randomly select one of the three baskets of food chosen by them as gifts.