Nudge and boosts to help individuals make better choices

We consider different versions of our consumer choice game, that differ regarding whether or not an instrument (a nudge or a boost) is implemented. Each version of the game corresponds to a treatment (i.e., intervention).

2023-06-09

2023-12-31

We will mainly analyze:
- subjects’ payoffs (that are given by the quantity of goods chosen taking into account their prices and the available budget);
- distance between the quantity chosen and the optimum value
- the time spent to take the decision, and;
- their degree of confidence (i.e., the extent to which they believe the chosen quantities are those that allow them to achieve the highest possible payoff).

At each repetition of the game, subjects decide quantities of good A and B (see below) they would like to buy, taking into account their prices and the available budget. Their choice of quantities directly gives them their payoff (first outcome variable). We are able to measure the distance from the optimal value (second outcome variable). In our setting, we control the time taken by subjects to take their decisions (third second outcome variable). Finally, once they have validated their choice, we ask subjects’ level of confidence on the fact that their choice allows them to reach the highest maximum payoff (fourth outcome variable).

To better interpret our results, after the main game of our experiment, we:
1) Implement the Cognitive Rationality Test (CRT) with revisited questions:
a) A golden bat and a golden ball cost 5000€ in total. The golden bat costs 4000€ more than the golden ball. How much does the golden bullet cost?
b) If it takes 10 machines 10 minutes to make 10 items, how long would it take 1000 machines to make 1000 items?
c) In a lake, there is a patch of water lilies. Every day, the plot doubles in size. If it takes 40 days for the plot to cover the entire lake, how many days would it take for the plot to cover a quarter of the lake?

2) A memory test consisting in memorizing 16 words in 1 minute. Then, in a list of 48 words, subjects have to select the 16 words they remember. Before selecting the words subjects remember, they perform the third task (see next point).

3) A computation test: subjects have to give the result of 20 calculus in one minute maximum.

Overall, these additional measures allow us to capture some potential heterogeneity among subjects regarding the effectiveness of our tools (nudge and boosts).
In particular, regarding the boosts, one may consider that those with a higher CRT score, who have a high score regarding the memory test, or regarding the computation test, are more likely to perform better under boost implementation (as well as after we remove it).

1) General setting:

Subjects play a consumer choice game in which we explain them that they have to choose quantities of (fictious) goods, good A and good B, they would like to buy, taking into account:
- the available budget;
- the prices (price A and price B) of each good.
The quantities they choose allow them to earn a payoff (expressed in Experimental Currency Units, ECU). The payoff function is based on a classic Cobb-Douglas function. Therefore, subject’s main objective is to solve an optimization problem at each round of the game.
In our game, subjects have no paper and pencil, nor calculator to make their decision (mobile phones will be switched off).
Subjects know that they cannot save money from their budget for other rounds. Moreover, if they spend more than the available budget, then subjects receive 0 ECU.

Subjects have to make their decision in a limited amount of time. The amount of time will be determined through a pilot study with around 25 subjects. The pilot study will use the same design with two minutes in each round to make their decision. To incite them to provide both quick, intuitive answers and more thoughtful ones, a round and a time within the two minutes will be drawn at the end of the experiment. This round and time will determine the participant's answer and the associated gain. It is therefore in the participants' interest to provide an answer as quickly as possible (as they receive no gain if they do not provide an answer before the randomly drawn time) and to keep thinking to eventually come up with a better answer before the two minutes are up.

The game is repeated, each round corresponding to different values of prices and budget.

Subjects select the quantities of good A and of good B using a cursor on sliders. To avoid any influence of the initial position of the cursor, subjects will have to first click on the screen to make appear the cursors. Then, they will be able to use them to select the quantities they want.

We consider three different phases of ten rounds:
- A first phase of ten round without any instrument being implemented.
- A second phase of ten rounds with an instrument (nudge or boost) or no instrument (control).
- A third phase of ten round without any instrument being implemented.

Before the start of the game, we will ask questions to measure subjects’ understanding of the game.

Finally, five rounds of the first phase are repeated in the second phase and, then, in the third one. This is another way to assess whether or not subjects learn over time.

2) Treatments:

We consider 4 different types of groups:
1) Control group: no instrument is implemented during the second phase of ten rounds.
2) Nudge treatment: a default option is implemented during the second phase of ten rounds. The optimal quantities of goods A and B are pre-selected for subjects (but they are not aware that the pre-selection corresponds to the optimal choice).
3) Boost treatment 1: at the beginning of the second phase (but before subjects play), subjects are shown a video to teach them the intuition to find the optimal quantities of goods A and B. Before the first round of the second phase, they can watch the video anytime they want. However, once they indicate that they understood, they can no longer watch it.
4) Boost treatment 2: during the second phase (at each round), subjects are shown the formula to compute the optimal quantities of goods A and B. This formula is no longer shown in the third phase.

We use the experimental platform to perform randomization. More precisely, participants register for a time slot for which they are available. Then, we randomize treatments at the session level. Participants can only participate once to our experiment.

At the session level for the treatment.

No

We will recruit around 25 subjects for the pilot experiment to determine the time limit of the final experiment.
We will consider a minimum of 60 subjects (independent observations in our case) in each treatment, for a (minimum) total of 60 x 4 treatments = 240 subjects for the final experiment.

We will recruit a minimum total of 25 + 240 = 265 subjects.

A minimum of 60 subjects per treatment.