Consent form for participation in this online experiment

Study Information

Thank you for your interest in our study. Please read the following information carefully. Your participation in this research is voluntary. It is your choice whether to participate or not. Please do not hesitate to write to us at aswath.chandrasekaran@student.uni-tuebingen.de if you have any further questions regarding the study.

1. Purpose

This study aims to investigate the role of LLM in decision making process and estimating risk at times of uncertainty.

2. Procedure

You will receive detailed instructions on how the task works, followed by a few questions to ensure you understand the task correctly. Next, you'll complete the full version of the task. Afterward, you'll complete questionnaires assessing numerical and risk taking preferences. Your performance in the main task will affect your bonus payment. The entire study will take approximately 15 - 20 minutes.

3. Compensation

You will receive £4 for your participation. You can also earn a bonus of up to £1.5 depending on your performance in the task.

4. Data protection

Once we have awarded you your bonus payment, your Prolific ID will be deleted from the study data. The experiment data will be stored under a randomly assigned participant number on a server at the University of Tübingen. This means that it will become impossible to link the study data to you. All data collected during this study will be treated confidentially and all individuals with access to your personal data are obliged to maintain data secrecy. We are not responsible for the handling of your personal data through the Prolific platform.
The study data may also be made publicly accessible via research databases or scientific publications (typically via the internet). This makes it possible for other researchers to check or replicate the results and enhances the quality of scientific research. The study data may also be used for new research questions going beyond the purposes of this particular study. The data is stored anonymously, treated confidentially, and used only for scientific purposes. You are free to withdraw, in which case recorded data will not be used (cf. point 6).

5. Voluntary participation

Your participation in this research is entirely voluntary. There will be no repercussions when canceling your participation.

6. Right to refuse or withdraw

You do not have to take part in this study if you do not wish to do so. You are free to withdraw your consent and do not need to provide a reason for your withdrawal. Participants who drop out of the study for any reason (e.g., technical issues) are eligible for full or partial compensation. Please contact aswath.chandrasekaran@student.uni-tuebingen.de and include a screenshot of the experiment website if possible. If you withdraw from the study before data collection is completed, your data will not be used. If you withdraw from the study after completing the study, but before bonus payments have been processed, you can request the deletion of your data. After the processing of bonus payments, your Prolific ID will be removed from the study data and it will be fully anonymized, making it not possible to link you with your data.

7. Risks

Your participation does not involve any physical or emotional risk to you. You are free to withdraw at any time.

8. Insurance coverage

Not necessary for participating in this study.

9. Responsible contact

If you have questions or if there are any problems, please contact aswath.chandrasekaran@student.uni-tuebingen.de. You can also use this email to request the withdrawal of your data.

Consent

By clicking on the button below, you agree to the following:

  • I have read and understood the conditions outlined above. I consent to participate in the study and agree to the collection, storage, and use of my data as described above.

Thank you for participating!

In this study, you will be asked to interpret results from medical diagnostic tests. You will review information about how accurate each test is and then make judgments about the likelihood of having the condition and what decision you would take. The study has two short sessions: one with only the test numbers and one with an AI-generated explanation. A performance based bonus will be awarded based on the accuracy of your estimates.

Welcome to the Diagnostic Dilemma!

You have received a medical test result that flagged you for a medical condition. As you know, no medical test is perfect and has to be interpreted carefully. In this experiment, you will weigh the evidence and decide what your next move will be. Each test scenario is summarized by three numbers:

  • Base Rate — The overall likelihood of the condition in the general population.
  • Sensitivity — The percentage of people with the condition who are correctly identified
    by the test (true positives).
  • Specificity — The percentage of people without the condition who are correctly cleared
    by the test (true negatives).

You have to weigh this evidence carefully to make an informed decision. You will complete two short sessions of decision-making scenarios:

  • Session 1 — Raw numbers only, no explanations.
  • Session 2 — Detailed AI report on what the base rate, sensitivity and specificity means

Click on Continue to see what you have to do in each scenario.


In each scenario, you will:

  1. Estimate the probability that you have the condition.
  2. Choose what to do next - seek treatment now or get a second opinion.

You will use a slider to indicate your probability estimate and answer two questions regarding decision on treatment. The slider is inactive at the start and becomes active once you click on it. Your bonus will depend on how well you estimate the probabilityand not on your decision. Your bonus will be be based on how well you estimate the probability of having the condition, not on your decision. You can earn a maximum bonus of £1.50.
Click on Continue to see an example of session 1:

Please fill out this Survey!



Please fill out this Survey!



🧠 Comprehension Check

1. What is the main difference between Session 1 and Session 2?






2. What will you do after reviewing each test result scenario?






3. What does the “base rate” represent in the test data?






4. What does “sensitivity” mean in the test data?






5. What does “specificity” mean in the test data?






Great! You have completed the instructions. Click on the button below to start the first session of the experiment.

Survey

We would like you to complete a short survey involving everyday numbers. These questions are about how people understand and work with simple information involving proportions and chances. Everyone approaches numbers differently, and we are interested in these differences to better understand how people interpret the medical test scenarios.

This short survey is for research purposes only and will not affect your performance bonus or how your answers in the main task are evaluated.

Click on the button below to begin the short survey.

Everyday Decisions Survey

People often see some risk in situations with uncertain outcomes and possible negative consequences. We are interested in your gut-level judgments about everyday situations and behaviors.

For each statement, you will answer TWO questions:

  1. Likelihood: How likely is it that you would engage in the described activity or behavior if you were in that situation?
    Scale: Extremely Unlikely → Extremely Likely
  2. Risk perception: How risky do you perceive this situation or behavior to be?
    Scale: Not at all Risky → Extremely Risky

Please move both sliders for each item before continuing to the next one.

Great! You are done with the instructions. Press Continue to start the main Experiment

You have completed the first session. Now Click on continue to start the second session! Remember that you will receive a bonus payment on how well you estimate the probability of having the condition, not on your decision. Press Next to go through an example

Click on Continue to start the second session of the experiment.

Your test results flagged you for the condition ABC


The probability of ABC is 10% (10 out of 100) for a person in the population. If a person has ABC, the probability that the test will return a positive result is 90% (90 out of 100). If a person does not have ABC, the probability that the test will return a negative result is 20% (20 out of 100)

🤖
AI Diagnostic Assistant
Online
AI
Hello! I'm your AI diagnostic assistant. I'm here to help you navigate through the numeracy tasks and answer any questions about probability, statistics, or diagnostic reasoning. How can I assist you today?
U
We have the following data from the test result in general: The base rate of the test: 0.31 The sensitivity of the test: 0.84 The specificity of the test: 0.69
AI
In a group of 100 people, 55 are expected to have the condition. With a sensitivity of 44%, 24 of these 55 individuals are correctly identified (true positives), while 31 are missed (false negatives). Among the 45 people without the condition, a specificity of 80% means 36 are correctly not flagged, and 9 are incorrectly flagged (false positives). Consider how the high base rate contrasts with the test’s limited ability to detect true cases and its moderate ability to avoid false alarms.
What do you think is the probability that you have the condition? Click on the slider!
%

Get second opinion Seek treatment

Please fill out this Survey!




Please fill out this Survey!




Thank you for participating!

Your total bonus is

Click on the link below to complete your submission:
Submit on Prolific

Trust and AI Experience Survey

This survey is designed to gather your experiences and perceptions regarding AI tools used. Your responses will help us understand how AI influences decision-making and trust in medical contexts. The answers won't affect your bonus payment

1. How experienced are you with using AI tools?

2. Did you use AI tools to complete the experiment?


3. Do you trust AI answers?