Experimental Design
We conduct an online survey experiment to study the effect of recall-based simulations of the future on how individuals discount future rewards
Overview of Experimental Design
We conduct an online survey experiment on individual decision-making. The main features of our experimental setup are (i) multiple pairwise intertemporal choices that participants make after (ii) pre-choice tasks.
In each intertemporal choice, subjects choose between two Amazon gift cards, gift card A and gift card B. Each gift card consists of two components: an observable value x; a date t during which the gift card would be made available (via email) to subjects.
In all choices, the date of gift card A, t_a=today, and the date of gift card B, t_b= in 4 weeks from now, remain constant. Between choices, we vary only the value of gift card B, x_b, to increase in steps of 1 Euro per choice.
In each choice, we have x_a<x_b and t_a<t_b: gift card A is always the sooner-smaller alternative and gift card B is always the larger-later alternative.
Between three treatments, the intertemporal choices are exactly the same. But we vary the pre-choice tasks, as we discuss below.
In addition, we elicit:
- whether subjects engaged in imagining consuming the voucher in 4 weeks from now
- how vivid subjects’ imagination of consuming the voucher in 4 weeks from now was
- whether the pre-choice task helped them to imagine consuming the voucher in 4 weeks from now
- subjects’ willingness to pay today for receiving a voucher of value x_a today
- subjects’ willingness to pay in 4 weeks from now for a receiving a voucher of value x_a in 4 weeks from now
- subjects’ estimated certainty over both of their willingness to pay measures
- subjects’ current emotional states (valence and arousal)
- subjects’ age, gender and regional location in Germany
Treatments
Our treatments differ only in the pre-choice tasks subjects face. In treatment 1, subjects first face a list of online shopping activities and are asked to indicate whether they have done these online shopping activities in the past and whether they will do them in 4 weeks from now. Subjects are then asked to recall aspects of their previous online shopping activities. Subjects are then asked to imagine doing online shopping in 4 weeks from now.
In treatment 2, subjects first face a list of activities that are not directly related to online shopping and are asked to indicate whether they have done these activities in the past and whether they will do them in 4 weeks from now. Subjects are then asked to recall a specific episode of one of the activities from the list of activities unrelated to online shopping. Subjects are then asked to imagine doing that activity in 4 weeks from now.
In treatment 3, subjects first estimate the weather in several foreign countries in 4 weeks from now. Subjects are then shown abstract paintings and are asked to write what they see. Subjects are then asked to imagine how their day in 4 weeks from now will be like.
Outcome variables
Our primary outcome is subjects’ average intertemporal choice.
Our secondary outcomes are:
- how vivid subjects’ imagination of consuming the voucher in 4 weeks from now was
- whether the pre-choice task helped them to imagine consuming the voucher in 4 weeks from now
- subjects’ willingness to pay today for receiving a voucher of value x_b today
- subjects’ willingness to pay in 4 weeks from now for a receiving a voucher of value x_b in 4 weeks from now
- subjects’ estimated certainty over both of their willingness to pay measures
Hypothesis and Tests
We test our prediction that subjects make more patient intertemporal choices in Treatment 1 than in Treatment 2 and that subjects record greater engagement with future simulation and greater vividness of future simulation in Treatment 1 than in Treatment 2.
We conduct Treatment 3 and elicit the remaining outcomes mainly for explorative purposes. However, we do expect that subjects’ patience in Treatment 3 generally falls between subjects’ patience in treatments 2 and 3, since we remove from Treatment 3 a recall task that, compared to no primed recall, is likely to boost consumption-relevant future simulation in Treatment 1 and to interfere with consumption-relevant future simulation in Treatment 2.
Exclusion criteria
We consider two samples. First, we consider all subjects who participated in our experiment. Second, we consider subjects who passed our attention checks, made consistent choices (ie whenever they were patient for some value of x_b they are not impatient for any larger values of x_b, or whenever they were impatient for some value of x_b they are not patient for any smaller values of x_b), and spend a reasonable amount of time completing the experiment.
The attention check refers to the following task each subject performs during the experiment. On each decision screen, we ask subjects to type the values of x_a and x_b of the current choice options into respective text fields. Subjects can type into these text fields anything. Subjects pass our attention check if they state the actual values of x_a and x_b of the current choice options. Subjects who make a mistake for at least one of these values fail our attention check.
Randomization and sample size
Subjects are randomly assigned to one of three (between-subjects) treatments. We aim for 1200-1300 subjects per treatment (depending on availability).