Human Behavior Science Project
Introduction
We would like to understand how human can distort probability information under risky decisions. According to past research, probability distortion occurs when people make decisions. Different ways of obtaining probability message will result in different distortion patterns.
This experiment is designed through lottery selection. If the information of probability and money are told directly to the subjects, people tend to overweight big-probabilities and underweight small-probability events. When probability information is learned through experience, the pattern of their behaviors on probability distortion is completely different.
In this investigation, we will conduct different experiments in order to estimate subjects’ probability weighting function (PWF) while they conduct decision making process.
Motivation
Classical economics believed that human make rational choices and should choose options with higher expected values. Afterwards, psychologists discovered that people’s decision making are not rational, they will distort the probability information.
According to Kahneman and Tversky’s prospect theory, it is mentioned that the characteristics exhibited by individual’s decision making under risk are inconsistent with the basic principles of Von Neumann-Morgenstern’s utility theory.
For example, when we provide two groups of lottery:
Although the expected value of A is larger than B, most of the subjects still choose the latter one. From this experiment, we can tell that people do not follow the classic economic theory. This is the phenomenon of probability distortion.
However, the results of the study from Hertwig were in contrast to Kahneman and Tversky’s perspective theory. They also provided two groups of lottery to their subjects, but this time they disclosed zero information about the probability and money:
The only way to discover the information back-end is to try and error. When subject selects A card, he will see $4 or $0, when subject selects B card, he will always see $3. In the experiment, subjects are allowed to try as much as they like to obtain the information. Then they could make their final decision to realize one of the lottery.
Hertwig found that when subjects learned the probabilities through their own experiences, their behavior were contrary to the traditional approach with directly notifying the information to subjects.
In the past, decision from experience experiments tell subjects the amount of money, and then allow subjects to learn the probability via repetitive drawing. However, we were wondering if subjects were affected by the amount of money or the probability itself. Thus, we would like to understand in the premise of not providing the information of money, do subjects still have distortions on probabilities?
Purpose
Understand when human conduct decision under risks, is the probability distortion phenomenon affected by money? Moreover, how the amount of experience affects their behavior of decision?
Research Questions
Related experiments on probability distortion under risky decision makings are roughly divided into two categories:
- Directly inform subjects the probability, and then estimate their probability distortion circumstances.
- Don’t tell subjects the probability message, but offering them to learn the probabilities by drawing the cards repetitively.
From the first category, most of the subjects amplify the feelings of probability when it is low, but decrease their feelings when the probability is high. This resulted in a distorted weighting function comparing to the real probability. However, in the second type of experiment, the subjects’ behavior is converse.
Based on these research foundations, we added some other factors in the experiment:
- We would like to know whether human distorted the probability itself or be influenced by the impact of money.
We designed this experiment by separating the information of probability and money. Subjects only learn the odds in the first stage, then they conduct lottery selection based on their learning.
2. We are curious about how experience affects decision making. Does different learning experience results in a different probability distortion?
We divided subjects into two groups. In the first stage, group one is provided with less learning opportunity, while group two is provided with sufficient learning opportunity. The second stage remains the same.
Expected results
In our second experiment, we believe by given more learning experience, the estimation of probability will approach the real probability. This will provide them more confidence in the second stage of lottery selection.
For the subjects having less learning experience, we predict their probability weighting function (PWF) will be a description type because of uncertainties of the real odds. Vice versa, we believe the other group with more learning opportunities will have an experienced type of PWF.
Preliminary results
It showed that we made an proper initial hypothesis: subjects with 10 times of learning have an estimated probability of being closer than those with 30 times of learning experience.
Regardless of how, one of the groups’ probability weighted function is different from our prediction. Both of the groups showed a “description type” of distorted shape instead of showing an “experienced type”.
Discussions
Why does the first group results in an experienced type of PWC as well? One of the reason according to our inference is: learning the odds first and then making lottery selection based on known odds is a bit like an alternative of telling subjects the probability directly. This may be the core reason why our experiment is more similar to Kahneman and Tversky’s.
Future Prospects
In the literature that explored empirical decision-making, it was rarely discussed why the distortion of probability under experience type is contrary to the outcome of descriptive type. In our experiment, we separate the information of probability and money, trying to clarify whether people are distorted in the decision-making process by the probability message itself or by the money. In addition, we also provide subjects with with different amounts and probabilities of lottery selection. Through subjects’ selection, the distortion pattern of their probability weight function is estimated.
Through the results of the experiment, we found that when referring to “experienced decision-making”, it is not so rigid and straightforward to determine that the distorted form of probability is necessarily to reduce the small probability and to enlarge the middle and big probability. When people make decisions, they do not follow the traditional economists expectation by choosing higher expectations. In the process of decision-making, there are many factors that affect people’s behavior. That’s why we kept wondering which factors lead to people’s irrationality, therefore has so far been a very interesting and worthwhile topic.
Reference
- 《Thinking, Fast and Slow》 — Daniel Kahneman
- “The description–experience gap in risky choice” written by Ralph Hertwig and Ido Erev
- “The Description–Experience Gap in Risky Choice: The Role of Sample Size and Experienced Probabilities” written by Robin Hau, Timothy J. Pleska,Jurgem Kiefer and Ralph Hertwig
- “Prospect Theory: An Analysis of Decision under Risk” written by Daniel Kahneman and Amos Tversky
- “Are Probabilities Overweighted or Underweighted When Rare Outcomes Are Experienced (Rarely)?” written by Christoph Ungemach, Nick Chater and Neil Stewart
- “Advance in prospect theory: cumulative representation of uncertainty” written by Daniel Kahneman and Amos Tversky(1992)
