Evaluating Non-EU Models Michael H. Birnbaum Fullerton, California, USA.

Slides:



Advertisements
Similar presentations
Paradoxes in Decision Making With a Solution. Lottery 1 $3000 S1 $4000 $0 80% 20% R1 80%20%
Advertisements

New Paradoxes of Risky Decision Making that Refute Prospect Theories Michael H. Birnbaum Fullerton, California, USA.
Among those who cycle most have no regrets Michael H. Birnbaum Decision Research Center, Fullerton.
Science of JDM as an Efficient Game of Mastermind Michael H. Birnbaum California State University, Fullerton Bonn, July 26, 2013.
This Pump Sucks: Testing Transitivity with Individual Data Michael H. Birnbaum and Jeffrey P. Bahra California State University, Fullerton.
1 Upper Cumulative Independence Michael H. Birnbaum California State University, Fullerton.
Components of Source Credibility Michael H. Birnbaum Fullerton, California, USA.
1 Lower Distribution Independence Michael H. Birnbaum California State University, Fullerton.
True and Error Models of Response Variation in Judgment and Decision Tasks Michael H. Birnbaum.
Who are these People Who Violate Stochastic Dominance, Anyway? What, if anything, are they thinking? Michael H. Birnbaum California State University, Fullerton.
1 st lecture Probabilities and Prospect Theory. Probabilities In a text over 10 standard novel-pages, how many 7-letter words are of the form: 1._ _ _.
Testing Lexicographic Semi- Order Models: Generalizing the Priority Heuristic Michael H. Birnbaum California State University, Fullerton.
Testing Heuristic Models of Risky Decision Making Michael H. Birnbaum California State University, Fullerton.
1 A Brief History of Descriptive Theories of Decision Making Kiel, June 9, 2005 Michael H. Birnbaum California State University, Fullerton.
Some New Approaches to Old Problems: Behavioral Models of Preference Michael H. Birnbaum California State University, Fullerton.
1 Distribution Independence Michael H. Birnbaum California State University, Fullerton.
Resampling techniques Why resampling? Jacknife Cross-validation Bootstrap Examples of application of bootstrap.
1 Upper Tail Independence Michael H. Birnbaum California State University, Fullerton.
458 Model Uncertainty and Model Selection Fish 458, Lecture 13.
Testing Models of Stochastic Dominance Violations Michael H. Birnbaum Decision Research Center California State University, Fullerton.
1 Upper Distribution Independence Michael H. Birnbaum California State University, Fullerton.
Ten “New Paradoxes” Refute Cumulative Prospect Theory of Risky Decision Making Michael H. Birnbaum Decision Research Center California State University,
Violations of Stochastic Dominance Michael H. Birnbaum California State University, Fullerton.
Testing Critical Properties of Models of Risky Decision Making Michael H. Birnbaum Fullerton, California, USA Sept. 13, 2007 Luxembourg.
Ten “New Paradoxes” Refute Cumulative Prospect Theory of Risky Decision Making Michael H. Birnbaum Decision Research Center California State University,
New Paradoxes of Risky Decision Making that Refute Prospect Theories Michael H. Birnbaum Fullerton, California, USA.
AN INTRODUCTION TO PORTFOLIO MANAGEMENT
1 The Case Against Prospect Theories of Risky Decision Making Michael H. Birnbaum California State University, Fullerton.
Testing Transitivity (and other Properties) Using a True and Error Model Michael H. Birnbaum.
Web-Based Program of Research on Risky Decision Making Michael H. Birnbaum California State University, Fullerton.
Web-Based Program of Research on Risky Decision Making Michael H. Birnbaum California State University, Fullerton.
1 A Brief History of Descriptive Theories of Decision Making: Lecture 2: SWU and PT Kiel, June 10, 2005 Michael H. Birnbaum California State University,
1 Gain-Loss Separability and Reflection In memory of Ward Edwards Michael H. Birnbaum California State University, Fullerton.
I’m not overweight It just needs redistribution Michael H. Birnbaum California State University, Fullerton.
1 Ten “New Paradoxes” of Risky Decision Making Michael H. Birnbaum Decision Research Center California State University, Fullerton.
PSY 1950 Confidence and Power December, Requisite Quote “The picturing of data allows us to be sensitive not only to the multiple hypotheses that.
1 Gain-Loss Separability Michael H. Birnbaum California State University, Fullerton.
Is there Some Format in Which CPT Violations are Attenuated? Michael H. Birnbaum Decision Research Center California State University, Fullerton.
1 Lower Cumulative Independence Michael H. Birnbaum California State University, Fullerton.
Stochastic Dominance Michael H. Birnbaum Decision Research Center California State University, Fullerton.
Web-Based Program of Research on Risky Decision Making Michael H. Birnbaum California State University, Fullerton.
Testing Transitivity with Individual Data Michael H. Birnbaum and Jeffrey P. Bahra California State University, Fullerton.
1 Restricted Branch Independence Michael H. Birnbaum California State University, Fullerton.
AN INTRODUCTION TO PORTFOLIO MANAGEMENT
Descriptive Statistics
Presidential Address: A Program of Web-Based Research on Decision Making Michael H. Birnbaum SCiP, St. Louis, MO November 18, 2010.
Behavior in the loss domain : an experiment using the probability trade-off consistency condition Olivier L’Haridon GRID, ESTP-ENSAM.
A Perfect Cocktail Recipe: Mixing Decision Theories and Models of Stochastic Choice Ganna Pogrebna June 29, 2007 Blavatskyy, Pavlo and Ganna Pogrebna (2007)
STOCHASTIC DOMINANCE APPROACH TO PORTFOLIO OPTIMIZATION Nesrin Alptekin Anadolu University, TURKEY.
Decision making Making decisions Optimal decisions Violations of rationality.
Agata Michalaszek Warsaw School of Social Psychology Information search patterns in risk judgment and in risky choices.
Population All members of a set which have a given characteristic. Population Data Data associated with a certain population. Population Parameter A measure.
Health State Unable to perform some tasks at home and/or at work Able to perform all self care activities (eating, bathing, dressing) albeit with some.
Chapter 9 Power. Decisions A null hypothesis significance test tells us the probability of obtaining our results when the null hypothesis is true p(Results|H.
VI. Evaluate Model Fit Basic questions that modelers must address are: How well does the model fit the data? Do changes to a model, such as reparameterization,
Stochastic choice under risk Pavlo Blavatskyy June 24, 2006.
Experiments on Risk Taking and Evaluation Periods Misread as Evidence of Myopic Loss Aversion Ganna Pogrebna June 30, 2007 Experiments on Risk Taking and.
A Stochastic Expected Utility Theory Pavlo R. Blavatskyy June 2007.
Lecture 16 Section 8.1 Objectives: Testing Statistical Hypotheses − Stating hypotheses statements − Type I and II errors − Conducting a hypothesis test.
Ellsberg’s paradoxes: Problems for rank- dependent utility explanations Cherng-Horng Lan & Nigel Harvey Department of Psychology University College London.
Buying and Selling Prices under Risk, Ambiguity and Conflict Michael Smithson The Australian National University Paul D. Campbell Australian Bureau of.
Testing Transitivity with a True and Error Model Michael H. Birnbaum California State University, Fullerton.
We report an empirical study of buying and selling prices for three kinds of gambles: Risky (with known probabilities), Ambiguous (with lower and upper.
Axiomatic Theory of Probabilistic Decision Making under Risk Pavlo R. Blavatskyy University of Zurich April 21st, 2007.
Allais Paradox, Ellsberg Paradox, and the Common Consequence Principle Then: Introduction to Prospect Theory Psychology 466: Judgment & Decision Making.
Can a Dominatrix Make My Pump Work? Michael H. Birnbaum CSUF Decision Research Center.
Data Mining Practical Machine Learning Tools and Techniques By I. H. Witten, E. Frank and M. A. Hall Chapter 5: Credibility: Evaluating What’s Been Learned.
Estimation & Hypothesis Testing for Two Population Parameters
Hypothesis Testing.
New Paradoxes of Risky Decision Making that Refute Prospect Theories
Presentation transcript:

Evaluating Non-EU Models Michael H. Birnbaum Fullerton, California, USA

Outline This talk will review tests between Cumulative Prospect Theory (CPT) and Transfer of Attention eXchange (TAX) models. Emphasis will be on experimental design; i.e., how we select the choices we present to the participants. How not to design a study to contrast with how one should devise diagnostic tests.

Cumulative Prospect Theory/ Rank-Dependent Utility (RDU)

Nested Models

Testing Nested Models Because EV is a special case of EU, there is no way to refute EU in favor of EV. Because EU is a special case of CPT, there is no way to refute CPT in favor of EU. We can do significance tests and cross- validation. Are deviations significant? Do we improve prediction by estimating additional parameters (Cross-validation)? (It can easily occur that CPT fits significantly better but does worse than EU on cross-validation.)

Indices of Fit have little Value in Comparing Models Indices of fit such as percentage of correct predictions or correlations between theory and data are often insensitive and can be misleading when comparing non-nested models. In particular, problems of measurement, parameters, functional forms, and “error” can make a worse model achieve higher values of the index.

Individual Differences If some individuals are best fit by EV, some by EU, and some by CPT, we would say CPT is the “best” model because all participants can be fit by the same model. But with non-nested models with “errors” it is likely that some individuals will appear “best” fit by a “wrong” model.

“Prior” TAX Model Assumptions:

TAX Parameters For 0 < x < $150 u(x) = x Gives a decent approximation. Risk aversion produced by  

Non-nested Models Special TAX and CPT are both special cases of a more general rank- affected configural weight model, and both have EU as a special case, but neither of these models is nested in the other. Both can account for Allais paradoxes but do so in different ways.

How not to test among the models Choices of form: (x, p; y, q; z) versus (x, p; y’, q; z) EV, EU, CPT, and TAX as well as other models all agree for such choices. Furthermore, picking x, y, z, y’, p, and q randomly will not help.

Non-nested Models

CPT and TAX nearly identical inside the prob. simplex

How not to test non-EU models Tests of Allais types 1, 2, 3 do not distinguish TAX and CPT. No point in fitting these models to such non-diagnostic data. Choosing random levels of the gamble features does not add anything.

Testing CPT Coalescing Stochastic Dominance Lower Cum. Independence Upper Cumulative Independence Upper Tail Independence Gain-Loss Separability TAX:Violations of:

Testing TAX Model 4-Distribution Independence 3-Lower Distribution Independence 3-2 Lower Distribution Independence 3-Upper Distribution Independence CPT: Violations of:

Allais Paradox 80% prefer R = ($100,0.1;$7) over S = ($50, 0.15; $7) 20% prefer R’ = ($100, 0.9; $7) over S’ = ($100, 0.8; $50) This reversal violates “Sure Thing” Axiom. Due to violation of coalescing, restricted branch independence, or transitivity?

Decision Theories and Allais Paradox Branch Independence CoalescingSatisfiedViolated SatisfiedEU, CPT* OPT* RDU, CPT* ViolatedSWU, OPT*RAM, TAX, GDU

Stochastic Dominance This choice does test between CPT and TAX (x, p; y, q; z) vs. (x, p – q; y’, q; z) Note that this recipe uses 4 distinct values of consequences. It falls outside the probability simplex defined on three consequences.

Basic Assumptions Each choice in an experiment has a true choice probability, p, and an error rate, e. The error rate is estimated from (and is the “reason” given for) inconsistency of response to the same choice by same person over repetitions

One Choice, Two Repetitions AB A B

Solution for e The proportion of preference reversals between repetitions allows an estimate of e. Both off-diagonal entries should be equal, and are equal to:

Estimating e

Estimating p

Testing if p = 0

Ex: Stochastic Dominance 122 Undergrads: 59% repeated viols (BB) 28% Preference Reversals (AB or BA) Estimates: e = 0.19; p = Experts: 35% repeated violations 31% Reversals Estimates: e = 0.196; p = 0.50 Chi-Squared test reject H0: p < 0.4

Results: CPT makes wrong predictions for all 12 tests Can CPT be saved by using different participants? Not yet. Can CPT be saved by using different formats for presentation? More than a dozen formats have been tested. Violations of coalescing, stochastic dominance, lower and upper cumulative independence replicated with 14 different formats and thousands of participants.

Implications Results are quite clear: neither PT nor CPT are descriptive of risky decision making TAX correctly predicts the violations of CPT; several predictions made in advance of experiments. However, it might be a series of lucky coincidences that TAX has been successful. Perhaps some other theory would be more accurate than TAX. Luce and Marley working with GDU, a family of models that violate coalescing.