|
Field
Trial Title
|
Before
State-dependent mispredictions and intertemporal choices
|
After
Preference misperceptions and intertemporal choices
|
|
Field
Abstract
|
Before
This study investigates preference misprediction, especially due to projection bias (Loewenstein et al.,2003) and the lack of experience, in intertemporal choices. The objectives are two-fold: first, it aims to demonstrate the existence of a behavioral tendency to project current preferences into the future from a state-dependent valuation pattern; second, it aims to show how this state-dependent misprediction of preferences confounds the identification of time preferences from state-varying choices. To this end, I run an online experiment with multiple sessions in an environment with unpleasant effort tasks (e.g. Augenblick and Rabin, 2018), where participants at different stages of mandatory work can forgo payoffs to avoid (or equivalently, require compensation for) additional work at different point of time.
|
After
This project investigates how preference misperceptions confound the identification of time preferences from state-varying choices. To this end, I run an online longitudinal experiment with multiple sessions in a stylized environment with unpleasant effort tasks, where participants at different stages of mandatory work can forgo payoffs to avoid (or equivalently, require compensation for) additional work at a different point in time.
|
|
Field
Trial Start Date
|
Before
March 01, 2021
|
After
March 29, 2021
|
|
Field
Trial End Date
|
Before
March 31, 2021
|
After
June 30, 2021
|
|
Field
Last Published
|
Before
February 28, 2021 05:13 PM
|
After
March 28, 2021 01:10 PM
|
|
Field
Intervention (Public)
|
Before
4 between-subject treatments: Good state w/o experience; Bad state w/o experience; Good state with experience; Bad state with experience
|
After
4 between-subject treatments: Good-Good; Good-Bad; Bad-Good; Bad-Bad. Details are explained in the attachment.
|
|
Field
Intervention Start Date
|
Before
March 01, 2021
|
After
March 29, 2021
|
|
Field
Intervention End Date
|
Before
March 31, 2021
|
After
June 30, 2021
|
|
Field
Primary Outcomes (End Points)
|
Before
Willingness-to-accept for (or equivalently, willingness-to-pay to avoid) one additional task
|
After
Self-reported prediction of tiredness at the end of Part 1 and at end of Part 2 (elicited in Part 1);
WTA choices for one additional task (Choice 1 and Choice 2 elicited in Part 1)
|
|
Field
Primary Outcomes (Explanation)
|
Before
|
After
Details are explained in the attachment.
|
|
Field
Experimental Design (Public)
|
Before
The study aims to detect state dependence in an intertemporal setting, where the state is varied by different degrees of aversiveness. Motivated by the paradigm in Augenblick and Rabin (2018), the experimental setting consists of unpleasant effort tasks and valuation choices of these tasks on top. The decision state is varied by the cumulated fatigue/tiredness from the tasks at different stages of mandatory work and the particular task involved is to count the numbers of “0”s in a matrix (e.g. Abeler et al., 2011).
The main experiment consists of two sessions that are two days apart. Participants are required to complete a certain number of mandatory tasks in both sessions. The main choice variable of interest is participants’ willingness-to-accept for additional work at different sessions elicited in different states, either good state (not tired) or bad state (tired). These within-subject choices vary in time horizon as well as decision states, depending on the treatment. State-dependent mispredictions predict that these valuation choices will be smaller in Good state than in Bad state treatment.
|
After
I use an online longitudinal experiment to demonstrate the identification problem of time preferences by showing how variation in decision states produces state-dependent behaviors in different time horizons. Motivated by the paradigm in Augenblick and Rabin (2018), the experimental setting consists of unpleasant effort tasks and valuation choices of these tasks on top. The decision state is varied by the cumulated fatigue/tiredness from the tasks at different stages of mandatory work.
|
|
Field
Planned Number of Clusters
|
Before
200 participants
|
After
n/a
|
|
Field
Planned Number of Observations
|
Before
200 participants (To ensure that the intertemporal choice, Choice 2, is incentive-compatible through a BDM mechanism and to be able to collect Choice 3 in Session 2, the experiment will only implement Choice 2 with some probabilities in Session 1. The reason is that those whose Choice 2 was implemented will not make Choice 3 in Session 2 since the additional task in Session 2 has been settled according to Choice 2. Thus, I will only have a full sample when analyzing state dependence from between-subject variations. When analyzing the interaction with time preferences using within-subject variations, I will only have a subsample who have made both Choice 2 in Session 1 and Choice 3 in Session 2.)
|
After
450
|
|
Field
Sample size (or number of clusters) by treatment arms
|
Before
50 participants in Good state w/o experience;
50 participants in Bad state w/o experience;
50 participants in Good state with experience;
50 participants in Bad state with experience.
|
After
Treatments where Part 1 is in a good state (Good-Good; Good-Bad): 225 subjects in total. The decision state in Part 2 will be assigned randomly at the beginning of Part 2.
Treatments where Part 1 is in a bad state (Bad-Good; Bad-Bad): 225 subjects in total. The decision state in Part 2 will be assigned randomly at the beginning of Part 2.
|
|
Field
Power calculation: Minimum Detectable Effect Size for Main Outcomes
|
Before
|
After
Details are explained in the attachment.
|
|
Field
Keyword(s)
|
Before
Behavior, Lab
|
After
Behavior, Other
|
|
Field
Intervention (Hidden)
|
Before
Good state w/o experience: without experiencing tasks in a separate session, participants make evaluation choices about Session 1 and Session 2 after the first out of ten tasks in Session 1.
Bad state w/o experience: without experiencing tasks in a separate session, participants make evaluation choices about Session 1 and Session 2 after the ninth out of ten tasks in Session 1.
*Good state with experience: having experienced ten tasks in a separate session before Session 1, participants make evaluation choices about Session 1 and Session 2 after the first out of ten tasks in Session 1.
*Bad state with experience: having experienced ten tasks in a separate session before Session 1, participants make evaluation choices about Session 1 and Session 2 after the ninth out of ten tasks in Session 1.
* The planned analysis will focus on the comparison between Good state and Bad state in order to demonstrate state dependence. In particular, the main comparison will be made for treatments with and without experience separately. Treatments with experience are meant for isolating the role of experience/learning from state dependence thus providing a clean test of projection bias.
* The magnitude of state dependence in treatments without experience and treatments with experience can differ for many other reasons thus are not entirely comparable. Thus, the quantitative exercise such as calculating the percentage role of a certain channel from the difference in magnitude should only be taken with a grain of salt. More emphasis should be put on the qualitative existence (or non-existence) of projection bias.
|
After
|
|
Field
Secondary Outcomes (End Points)
|
Before
Self-reported prediction of tiredness
Self-reported and hypothetical measures of patience;
Self-reported behaviors that are typically attributed to time preferences (saving, procrastination in coursework, and vaccine take-up)
|
After
Self-reported prediction of tiredness at the end of Part 2 (elicited in Part 2);
WTA choices for one additional task (Choice 3 elicited in Part 2)
|
|
Field
Secondary Outcomes (Explanation)
|
Before
|
After
Details are explained in the attachment.
|