1 / 19

70-386 Behavioral Decision Making

70-386 Behavioral Decision Making. Lecture 5: Even More Biases. Administrative. Paper presentations First paper presentation is Thursday Second is next Tuesday Quiz 1 results Scores soon Troubling issue: making up mechanisms resolve overconfidence

bendek
Télécharger la présentation

70-386 Behavioral Decision Making

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. 70-386 Behavioral Decision Making Lecture 5: Even More Biases

  2. Administrative • Paper presentations • First paper presentation is Thursday • Second is next Tuesday • Quiz 1 results • Scores soon • Troubling issue: making up mechanisms resolve overconfidence • Just because it’s not quantitative doesn’t mean you get to make up an answer that sounds good.

  3. Last time • Bayes Theorem • Base-rate neglect • Gambler’s fallacy, hot hands, and other misconceptions of chance

  4. Regression to the Mean • We’re generally really bad at understanding the probabilistic nature of most random events. • It’s not that bad outcomes will follow good ones, or visa versa. Average outcomes follow either • Examples: aka the ‘sophomore jinx’ • Movie sequels are often worse than the great movies by which they are prompted • Athletes who appear on Sports Illustrated covers often then experience worse performance (the “SI curse”) • Punishment works; praise doesn’t Business applications? • Performance evaluations

  5. Conjunction Fallacy Linda is 31 years old, single, outspoken, and very smart. She majored in philosophy. As a student, she was deeply concerned with issues of discrimination and social justice, and she participated in antinuclear demonstrations. • Rank the following eight descriptions in order of probability (likelihood) that they describe Linda: • ___a. Linda is a teacher in elementary school. • ___b. Linda works in a bookstore and takes yoga classes. • ___c. Linda is active in the feminist movement. • ___d. Linda is a psychiatric social worker. • ___e. Linda is a member of the League of Women Voters. • ___f. Linda is a bank teller. • ___g. Linda is an insurance salesperson. • ___h. Linda is a bank teller who is active in the feminist movement.

  6. Conjunction Fallacy Linda is 31 years old, single, outspoken, and very smart. She majored in philosophy. As a student, she was deeply concerned with issues of discrimination and social justice, and she participated in antinuclear demonstrations. • Rank the following eight descriptions in order of probability (likelihood) that they describe Linda: • ___a. Linda is a teacher in elementary school. • ___b. Linda works in a bookstore and takes yoga classes. • ___c. Linda is active in the feminist movement. • ___d. Linda is a psychiatric social worker. • ___e. Linda is a member of the League of Women Voters. • ___f. Linda is a bank teller. • ___g. Linda is an insurance salesperson. • ___h. Linda is a bank teller who is active in the feminist movement.

  7. Conjunction Fallacy Most people rank 3 > 2. Actually, 100% of you did. • Linda is active in the feminist movement. • Linda is a bank teller. • Linda is a bank teller who is active in the feminist movement. Can’t be. Why? • P(A &B) ≤P(A) and P(A & B) ≤P(B) • Being a subset (bank teller AND feminist) can’t be larger than the set that includes it (bank teller) • The conjunction of two events is always equal or less probable than the individual events • But…conjunction often provides or completes the “story” Implications? • People find it very difficult to reason about isolated events • People in business often reason by anecdote (e.g., case studies, “war stories”), but such reasoning is often grossly biased when it comes to communicating probabilistic information

  8. Representativeness Results of a recent survey of 74 Fortune 500 CEOs indicate that there may be a link between childhood pet ownership and future career success. Fully 94% of them had possessed a dog, or cat, or both as youngsters . . . . The respondents asserted that pet ownership had helped them develop positive character traits that make them good managers today: responsibility, empathy, generosity, and good communication skills.” -Management Focus, November 1984 • How about this? • “Fully 100% of the CEOs brushed their teeth as children…” • Why not make that assertion?

  9. Confirmation Trap • Suppose you had four cards in front of you and you wanted to test the following statement: ‘if a card has a vowel on one side, then it has an even number on the other side.’ Which cards would you turn over? • You said • E: 12/13 K: 2/13 4: 6/13 7:4/13 • Actually, you need E and 7 (2 of you were correct) E K 4 7

  10. Confirmation Trap • Biased information search • People inherently search for information to confirm their beliefs (or in this case a rule) • But you should search for disconfirming evidence too • A vowel andan odd number.

  11. Anchoring & Adjustment Is the height of the tallest building in Dubai greater or less than 300 meters? Greater thanLess than How tall is the tallest building in Dubai? ______ Is the height of the tallest building in Dubai greater or less than 1,300 meters? Greater thanLess than How tall is the tallest building in Dubai? ______

  12. Anchoring & Adjustment • When was the TajMahal completed? • We didn’t have enough data for this, but even random numbers can anchor people (your telephone number). • People use an anchor (sometimes set arbitrarily) as a starting point • They adjust in what they believe to be the correct direction. They just don’t move enough. • Business applications?

  13. 𝜆 ρ β θ

  14. Anchoring & Adjustment • Biased information search / estimation. • Estimate 8 * 7 * 6 * 5 * 4 * 3 * 2 * 1? • Median estimate of 2250 (KT 1974) • What is 1 * 2 * 3 * 4 * 5 * 6 * 7 * 8? • Median estimate of 512 (KT 1974)

  15. Anchoring and Adjustment Actual real estate agents were shown a house and given a 10-page packet of information in which only one number was varied for the experiment: the listing price. Although all the agents found the listing price to be too high, they anchored on this (arbitrary) value. The arbitrary listing price shifted their appraisals by more than $10,000! Additionally, only 1 in 10 agents mentioned the listing price as a factor in their judgment

  16. Conjunctive and Disjunctive Events Which of the following instances seems most likely? _______ Second most likely?________ • Drawing a red marble from a bag containing 50% red marbles and 50% white marbles. • Drawing a red marble seven times in succession, with replacement, from a bag containing 90% red marbles and 10% red marbles. • Drawing at least one red marble in seven tries, with replacement, from a bag containing 10% red marbles and 90% white marbles. • 6 of you said B-A-C • 3 said A-C-B • 1 said A-B-C  Most people say B-A-C Actual answer is C-A-B.

  17. Conjunctive and Disjunctive Events • People overestimate conjunctive events (“and”) • All subcomponents must occur for the event to occur • And underestimate disjunctive events (“or”). • Any of the subcomponents can occur for the event to occur. People aren’t intuitive statisticians. • So what? That’s why we take classes… • Business applications?

  18. Hindsight and Curse of Knowledge • Hindsight: • People think they knew things when they didn’t • Who does this? • Nearly every stock-picker and pundit. Everywhere. (really: everyone) • Curse of Knowledge • Once we know something, it’s really hard to image that we didn’t. • A main driver of communications failures

  19. Fundamental Attribution Error • Reading posted on the FAE and actor-observer bias tonight.

More Related