7.1 Causation, Blame, and Policy

Distinguishing singular causation (A caused B) from general causation (X tends to cause Y).

Useful Links

• Three Column Overview of the Week
• Lesson Slides (2023 Master)
• Website Page

Learning Goals

After this lesson, students should

1. Distinguish between singular and general causation.
2. Distinguish between the evidence needed to establish singular or general causation.
3. Identify different policy implications of singular or general causation.
4. Recognize cases where omission bias and status quo bias can influence decision making, even when this results in a worse outcome.

Definitions

• Singular Causation

  A causal relation between two specific events; A caused B.

• General Causation

  A causal relation between two variables; X causes Y.

• Omission Bias

  Omission bias is the tendency to favor an act of omission (inaction) over one of commission (action).

• Status Quo Bias

  A preference for the maintenance of a current or previous state of affairs ("status quo" is Latin for "state in which"), or a preference to not undertake any action to change this current or previous state.

Examples

• A randomized controlled trial for a drug may demonstrate its general therapeutic effect (general causation), e.g. taking this drug reduces the risks of this disease, but it cannot conclusively demonstrate a causal connection in any particular instance of a patient taking this drug (singular causation).
• One can claim that climate change causes extreme weather events (general causation), but not that any particular instance of wildfire was caused by climate change (singular causation).

Common Misconceptions

• Some students think that an RCT can demonstrate singular causation, whereas in fact it can only establish general causation.

Context

This course has so far only discussed general causation, which can be demonstrated through randomized controlled trials and to a weaker extent Hill's criteria. But personal, policy, and legal decisions often depend on singular causation as well. It also sometimes matters whether the causation is by commission or by omission. The famous Trolley dilemma is discussed.

Before

6.1 Correlation and Causation

RCTs only demonstrate general causation but not singular causation. Correlation is also a statistical relationship between two variables, rather than between two singular events.

6.2 Hill's Criteria

As a substitute for RCTs, Hill's criteria also only demonstrate general causation, but can be used in some cases as partial evidence for singular causation (especially plausible mechanism and temporal sequence).

After

8.1 Orders of Understanding

Any given effect is brought about by a complex combination of many causes (which may interact with each other), with varying degrees of influence on the outcome. This is true for both singular and general causation.

9.2 Biases

The omission bias is one explanation for the status quo bias, in that humans tend to prefer not actively changing the current situation or trend, even when it may be worse than the potential pitfalls of the new outcome.

Recommended Outline

Before Class

• Review PlayPosit and discussion questions and ask faculty, Gabriel, or Emlen any questions you have.
• (Optional) Prepare a presentation.

During Class

• (5 min) Come up with some fun way to assign the roles of spokesperson and notetaker (e.g. earliest birthday in the year, lives furthest from campus). Remind them of the responsibilities of these roles.
• (35 min) Guide the students through the singular and general causation discussion questions.
• (15 min) Discuss the trolley problem example.
• (25 min) Go through the life saving treatments discussion.

After Class

• Collect answers from notetakers for the forum / plenary.

Lesson Content

Singular and General Causation Discussion Questions

In small groups, have students discuss the following questions:

1. Which kind of causation can an RCT establish?

2. Suppose we conduct an RCT that determines salt causes heart disease. Suppose Sally is in our study, has a high level of salt, and contracts heart disease. Did Sally's salt intake cause her heart disease?

3. On January 4, 2022, a federal inquiry into the Dixie Fire of August 2021, the second largest wildfire in California history, determined the wildfire began when a tree came in contact with a power line operated by PG&E.
   1. Is this a statement of singular or general causation?

   2. It is well established that climate change increases the number and intensity of wildfires in California. Given this known fact, why did the federal investigation not immediately conclude that climate change was the cause?

   3. PG&E has faced criminal charges from multiple California counties for the role of its equipment in causing wildfires. When determining guilt or innocence, does the legal system typically focus on singular or general causation?

   4. Assume you're in charge of creating policy to prevent California wildfires like the Dixie Fire from happening in the future. What are some policies that might have prevented the Dixie Fire? Does policy making typically focus on singular or general causation?

4. Why do you think it is so tempting to assume that each event has a singular person or entity to blame? When is this useful or not useful?

Commission and Omission Discussion Questions

Trolley Problem

The point of the trolley problem is to emphasize the effect of omission bias. In doing so, it shows the different weights society and individuals tend to put on acts of commission versus acts of omission.

If you deem it appropriate, you can show this clip from the show The Good Place.

In small groups, have students discuss the following questions:

1. Would you pull the lever to shift the trolley from the track with five people to the track with one (assuming whoever is on the track will be crushed)? 👀
2. Are you causing the death of 1 person if you do pull the lever?
3. Are you causing the death of 5 people if you don't pull the lever?
4. Is there a significant distinction between the deaths that result from you pulling the lever (commission) vs the deaths if you don't (omission)?

Life Saving Treatments

Bring the students back and have them share what they talked about. Then explain a hypothetical scenario where you are a doctor in a hospital where there are five sick people whose lives could be saved using the organs of one healthy person, who will die when their organs are taken. Have the students discuss the following questions:

1. Should the healthy organs of the one person be used to save the five people?
2. What if instead of organs, it was a drug that could save either one person or five other people? Would you choose to save one person or five people?
3. Why do you think we place more blame on "sins of commission" than on "sins of omission?" Is it (never/sometimes/always) rational?

Then introduce "omission bias," which is the human tendency to favor an act of omission (inaction) over one of commission (action). For example, people give more blame to bad consequences of action than bad consequences of inaction; people give more credit for action with good consequences than inaction with good consequences. It is one explanation of the status quo bias, in that people prefer not to actively change the current situation, even if the change is more likely to result in a good outcome. The omission bias itself is not necessarily rational or irrational.

Finally, provide this real world example. Here, "subjects are reluctant to vaccinate a (hypothetical) child when the vaccination itself can cause death, even when this is much less likely than death from the disease prevented."

Suppose there is an emergent disease in a population. If left untreated, 1 million people will die from it. A new vaccine is available with 100% effectiveness in preventing the disease. However, if the whole population were vaccinated, then an estimated 100 people would die as a result of adverse reaction to the vaccine.

1. Show of hands: Would you give this vaccine to your child?
2. Show of hands: Would you authorize the distribution of this vaccine if you were the FDA?
3. What's the largest number of deaths from the vaccine that you would tolerate and still give it to your child? (0, 1, 100, 1 thousand, 10 thousand, 100 thousand, 1 million, above 1 million) Authorize if you were the FDA?

---

⟨

Previous

6.2 Hill's Criteria

⟩

Next

7.2 Emergent Phenomena