Queuing Decisions

2024-03-11

2024-03-18

gamma_2, gamma_3

gamma_2 = Takes value 1 if the subject answers 'Urgent' in section 2 when assigned as 'Type N'

gamma_3 = Takes value 1 if the subject answers 'Urgent' in section 3 when assigned as 'Type N'
Note: for gamma_3, for control 1 and treatment 1, I will randomly select and keep only the number of observations which ensures that at least 90% of the samples consist of subjects who were truthful in section 1. Similarly, for control 1 and treatment 2, I will only keep the number of observations which ensures that 90% of the samples consist of subjects who were non-truthful in section 1.

gamma

gamma = Takes value 1 if the subject answers 'Urgent' in section 3 when assigned as 'Type N'

Exploratory Analysis:
I plan to use this variable to investigate if the belief about other's propensity for truth-telling also plays a role in determining behavioral response in this setting.

Online Survey:

I will conduct an online survey using prolific to gather data on individual's decisions regarding the truthful disclosure of private information when queuing for an appointment under different conditions. Subjects will be randomly assigned to one of the four following groups:

1.) Control 1: No Fee - Honest Group
2.) Control 2: No Fee - Dishonest Group
3.) Treatment 1: Priority Fee - Honest Group
4.) Treatment 2: Priority Fee - Dishonest Group

The survey will consist of three sections, each requiring subjects to make two decisions. In each section, subjects will first learn about the setup of the interaction, the two possible types of individuals that they could be assigned as and the corresponding payoff for each type. Then, they will randomly be assigned a type and will have to choose whether to honestly disclose their assigned type. Subjects will be required to make a decision once as each type, with the order of type allocation being randomized. The specific instructions and questions presented to participants will differ according to the control or treatment groups to which they have been assigned. More specifically:

- In section 1, all subjects will be told that they will receive $1.50 for answering Type A and $0 for answering Type B. They will then be randomly assigned a type and asked what type they are. They will answer twice, once as each type.

- In section 2, all subjects are told that they will simulate a queuing scenario with another randomly selected participant on the survey in which they are both trying to get a doctors appointment . They will be told that there are two types:
--- Type N (Non-Urgent): with a low valuation for appointment. This type will receive $3 for first appointment and $1.50 for second (plus $7)
--- Type U (Urgent): with a high valuation for appointment. This type will receive $10 for the first appointment and $5 for the second
They are also told that the first appointment will be given to the 'urgent' case. If both cases are the same type, then the assignment is random. Further, the participants in the treatment groups are informed that if they answer that their case is 'urgent' they will have to pay a priority fee of $0.75. They will then be randomly assigned a type and asked whether they would like to tell the receptionist at the doctor's office that their case is 'urgent' or 'non-urgent'. They will answer once as each type.

- In section 3, all subjects will be presented with a modified scenario and asked to answer the questions from section 2 again. More specifically:
--- Subjects in Control 1 and Treatment 1 will be told that they have been randomly assigned to a group of 10 individuals and that the decisions they make in the section will give them extra payment only in the event that 9 out of the 10 subjects in their group had been truthful in their responses in Section 1.
--- Subjects in Control 2 and Treatment 2 will be told that they have been randomly assigned to a group of 10 individuals and that the decisions they make in the section will give them extra payment only in the event that 9 out of the 10 subjects in their group had not been truthful in their responses in Section 1.
The subjects will answer once as each type.


Queuing Game and Payment:

All subjects receive a fixed base payment of $2 for participating in the survey. Each subject will also receive additional payment for one of the first four decisions they make on the survey.

If one of the responses for section 1 get selected for compensation, the participants will receive $1.50 if they have answered Type A and $0 if they have answered Type B.

If one of the responses for section 2 get selected for compensation, subjects' payments are determined through a queuing game, where:
- Each subject is randomly paired with another subject from their assigned treatment or control group.
- The other subject's response as Type N is selected with 75% probability and Type U with 25% probability.
- If the subject has answered Type U while their partner has answered Type N, the subject is awarded the first appointment and the corresponding payment.
- Conversely, if the subject has answered Type N while their partner has answered Type U, the subject is awarded the second appointment and it's corresponding payment.
- If both participants have the same answer, the first appointment (and payment) is randomly assigned.

For section 3, subjects become eligible for additional payment only if they get assigned to a 10 subject group that meets specific eligibility criteria. Initially, participants will be randomly divided into groups of 10, each within their designated treatment or control group. The compensation rules from section 2 apply here as well. However, the key difference is that for this section, subjects will be paired exclusively with another subject from their own 10 subject group.

Randomization done by computer. Subjects are randomly allocated into the priority fee treatment group or the no fee control group at the end of section 1. Then at the end of section 2, the subjects in the no fee control groups are randomly allocated into either control group 1 or 2. Similarly, subjects in the priority fee treatment group are randomly allocated into either treatment group 1 or 2.

Individual

No

700 subjects

700 subjects

The subjects will be divided equally across the four different groups (i.e. approximately 175 subjects in each treatment arm).

I conducted a simulation based power analysis (STATA code is available upon request). I adapted a latent variable approach and assumed that individuals have an underlying propensity for truth-telling which follows a standard normal distribution. I also assume that non-truthfulness in section 1, when assigned as a Type B individual, explains 55% of non-truthful behavior in the subsequent stages. Furthermore, I assumed that around half of the participants are likely to be non-truthful in the absence of a priority fee. Further, I assume that priority fee reduces this inclination by a little over half, i.e. I study an effect size of 56%. I will test the three following main hypotheses: 1.) Implementing a priority fee will reduce the proportion of lower-need individuals who are non-truthful: I will preform a Wilcoxan ranksum test on the variable gamma_2. For this between-subject test I have 100% power to find a significant effect at the 5% level. 2.) The priority fee will decrease the proportion of low-need individuals who are non-truthful in both the treatment groups: I will preform a Dunn test on the variable gamma_3. For this test, at the 5% level, I have 96.4% power to find a significant effect for treatment and control groups 1 and 100% power to find a significant effect for treatment and control groups 2. 3.) The priority fee will have a larger effect in the group 2 compared to group 1: I will preform a regression analysis where I examine the interaction effects between the treatment and control groups. For this test, I have 87.3% power at the 5% level to find a significant effect.