12.1: Wisdom of Crowds & Herd Thinking

April 12, 2021

Learning Objectives

1. Recognize when groups are likely to make good decisions (wisdom of crowds), and when groups are likely to make bad decisions (herd thinking).

​

• Identify features of existing group decision making structures which could be improved.

​

Agenda

1. Polls & Concept Review
2. Small Group Discussion
3. Main Group Discussion
4. Exit Survey

Fill out Google form, link in chat

You’ll have 3 minutes to fill out the Google form, just give your best guesses for each answer. We’ll come back to these responses in a bit.

​

Zoom Poll: Which artist created this painting?

1. Armand Guillamin
2. Eugene Boudin
3. Paul Cezanne

Zoom Poll: Which artist created this painting?

1. Armand Guillamin
2. Eugene Boudin
3. Paul Cezanne

​

Generally, most people will say Paul Cezanne because he is more well-known than Boudin or Guillamin. This is an example of herd thinking leading to an incorrect answer, because the group shares a prior bias.

Herd Thinking

Sometimes groups make worse judgments than individuals. 

​

This happens when: 

• Judgments of individuals are influenced by the judgments of others, leading to groupthink and sometimes polarization. 
• Members of the group share biases, which can be exaggerated by discussion and cannot be decreased by averaging judgments.

​

Ex: "We can be pretty confident we are right, because all five members of our family agree that vaccines cause autism. Since all five of us think so, we must be right."

​

Google form answers

Have UGSI screenshare Google form response summary, correct answers below:

​

1. What is the heaviest recorded weight for a cat as of 2013 (in pounds)? 🡪 922 lbs

2. What was the average weight for American men in 1960 (in pounds)? 🡪 166 lbs

3. What is the heaviest recorded weight for a rabbit as of 2013 (in pounds)? 🡪 55 lbs

4. What was the average weight for American women in 2010 (in pounds)? 🡪 163 lbs

​

​

Generally, when groups are asked for estimates on values for which they have no prior biases, they will collectively gravitate closer to the correct answer than an individual estimate may be able to. This is an example of the wisdom of crowds, where groups are able to make better answers than individuals.

Wisdom of Crowds

Sometimes groups make better judgments than individuals. 

​

This happens when: 

• Judgments are genuinely independent, preventing herd thinking.
• Members of the group do not share the same biases. 
• There are enough people in the group to balance out random biases or fluctuations (analogous to the need for an adequate sample size). 

​

Works especially well when estimating a quantity, where errors may be large but are not systematic. 

​

Ex: "We can get a pretty good estimate of the weight of this turkey by asking everyone in the family to write their guess privately on a piece of paper, and then averaging the answers."

​

​

Real-world Example: Jury Composition

A jury of 6 is constitutional (Williams v. Florida, 1970), but a jury of 5 is not (Ballew v. Georgia, 1978).

​

In a jury of 6, a decision must be unanimous (Burch v. Louisiana, 1979); in a Jury of 12, a majority of 9:3 or 10:2 can be constitutional for non-capital offenses (Johnson v. Louisiana and Apodaca v. Oregon, 1972).

​

​

Is a jury of 6 with 1 person who believes not guilty likely to lead to the same outcomes as a jury of 12 with 2 people who believe not guilty? What is the Supreme Court implicitly assuming about how group decision-making works?

​

​

In other words, are some juries more likely than others to come to innocent or guilty convictions just based on the psychology of group decision-making?

How does jury deliberation affect final decisions? (The optimistic answer.)

Kaplan & Miller (1978): Argued that deliberation would help correct juror biases, since final judgment would be a function of the average bias. However, it is also assumed that the evidence is emphasized in these deliberations (rather than the bias).

How does jury deliberation affect final decisions? (The pessimistic answer.)

Group polarization: group deliberation tends to exaggerate pre-existing group tendencies, whether they relate to political views, racial views, business policies, etc.

​

MacCoun (1984, 1990): Consistent with other work on group polarization, found that juries tended to overweight a completely irrelevant consideration: the defendant’s attractiveness.

How does jury deliberation affect final decisions? (The compromise.)

Kerr, Niedermeier, & Kaplan (1999): A possible reconciliation? Groups magnify biases when good sources of evidence are weak.

Summary of key points

• Wisdom of crowds: averaging responses from a large sample of independent guesses can often result in more accurate estimates. This works via error cancellation, and can therefore reduce statistical uncertainty, but not systematic sources of uncertainty that are common across individual people’s guesses.
• In some cases, group deliberation can result in worse decisions.
• When guesses are not independent, it can reduce wisdom of crowds effects.
• There’s no guarantee that the truth will win; often requires multiple advocates.
• Groups can polarize extreme positions, especially in the absence of unambiguous evidence one way or the other.

Small group discussion (15 mins)

1. Assign a spokesperson (whoever woke up earliest) and a notetaker (whoever woke up latest).
2. Can you think of a group you've worked with that engaged in groupthink/herd thinking? What happened?
3. Each person think of the best-functioning group they've worked with. What made that group stand out?
4. Suppose you are the chair for a senate committee on vaccines. There are only five people on your committee. What are two things you can do to try to reduce the possibility of herd thinking?
5. Notetaker: post a question or thought for faculty in the Ed discussion forum: https://edstem.org/us/courses/3936/discussion/358633

Exit Survey

Please fill out the following Exit Survey for a participation point, before you go!

​

https://forms.gle/PXecS6rfLtRMb4fAA

​

​

​

​

SUBMIT EXIT SURVEY

FIRST

LOG

OFF