9
2/8/2012 1 Chapter 11: Causal Explanation Critical Thinking, 10 th edition By Brooke Noel Moore and Richard Parker By PresenterMedia.com PowerPoint developed by Sarah Landsdown © 2011 PHYSICAL CAUSAL Using the physical characteristics or background to explain an event. Using behavior (psychology, poli sci, social sciences, etc.) to explain an event. BEHAVIORAL CAUSAL Types of Explanations Which one is which? Which one is which? 1. Why is the sky blue? 2. Why do people like Julie Roberts? 3. How did she become bitter? 4. What causes the rift between political parties? How to tell if an explanation is adequate? Circular Nontestable Testability Unnecessary Complexity If you can test out the predictions and they hold true, then the explanation is adequate If you repeat what you are meant to prove, isn’t adequate. Joe is poor because he doesn’t have any money. Just because you can’t test it doesn’t mean it’s not a good hypothesis Only abandon if untestable in principle. Adding in factors that are irrelevant or make the situation more complex. The food tasted horrible potentially because of too much salt or the stove was too hot.

PowerPointdeveloped bySarahLandsdown ©2011coursecontent.ntc.edu/economics/landsdown/cct/f2f/home/...2/8/2012 2 1. Be consistent 2. Not conflict with established fact or theory 3

  • Upload
    others

  • View
    1

  • Download
    0

Embed Size (px)

Citation preview

Page 1: PowerPointdeveloped bySarahLandsdown ©2011coursecontent.ntc.edu/economics/landsdown/cct/f2f/home/...2/8/2012 2 1. Be consistent 2. Not conflict with established fact or theory 3

2/8/2012

1

Chapter 11:  Causal Explanation

Critical Thinking, 10th editionBy Brooke Noel Moore and Richard Parker

By PresenterMedia.com

PowerPoint developed by Sarah Landsdown © 2011

PHYSICAL CAUSAL

• Using the physical characteristics or background to explain an event.

• Using behavior (psychology, polisci, social sciences, etc.) to explain an event.

BEHAVIORAL CAUSAL

•Types of Explanations

Which one is which?Which one is which?1. Why is the sky blue?2. Why do people like Julie Roberts?3. How did she become bitter?4. What causes the rift between political parties?

•How to tell if an explanation is adequate?

CircularNontestableTestabilityUnnecessary Complexity

• If you can test out the predictions and they hold true, then the explanation is adequate

• If you repeat what you are meant to prove, isn’t adequate.

• Joe is poor because he doesn’t have any money.

• Just because you can’t test it doesn’t mean it’s not a good hypothesis

• Only abandon if untestable in principle.

• Adding in factors 

that are irrelevant or 

make the situation more complex.

• The food tasted 

horrible potentially because of too much 

salt or the stove was 

too hot.

Page 2: PowerPointdeveloped bySarahLandsdown ©2011coursecontent.ntc.edu/economics/landsdown/cct/f2f/home/...2/8/2012 2 1. Be consistent 2. Not conflict with established fact or theory 3

2/8/2012

2

1. Be consistent

2. Not conflict with established fact or theory

3. Be testable

4. Not be circular

5. Avoid unnecessary assumptions or other unnecessary complexities

•Minimum adequate explanation requires…

complexities

•A casual explanation offered for further investigation or testing

•Hypothesis

•How to form a hypothesis?

•Method of Difference•Method of Agreement

•Casual Mechanisms and Background Knowledge

•The Best Diagnosis Method

Page 3: PowerPointdeveloped bySarahLandsdown ©2011coursecontent.ntc.edu/economics/landsdown/cct/f2f/home/...2/8/2012 2 1. Be consistent 2. Not conflict with established fact or theory 3

2/8/2012

3

•The Method of Difference

Eg. This stew turned out bad tasting this time…why?  You used a different type of ground chunk this time (generic vs. organic)

•The Method of Agreement

Eg. Your stairs get icy frequently at night, after the sun sets.  You find it is caused by your dryer vent when you do laundry.

•Don’t make the mistake!

• Just because there is a correlation, doesn’t mean it proves cause and effect!

• Cum hoc, ergo propter hoc – “with that, therefore because of that”

• Fallacy that correlation proves that one causes the other

• Eg.  When I do laundry, ice is going to be created.

• Post hoc, ergo propter hoc – “after that, therefore because of that”

• Fallacy that if something happens first it caused the following event.

• Eg.  Jumping up and down in your home caused an earthquake in China.

Page 4: PowerPointdeveloped bySarahLandsdown ©2011coursecontent.ntc.edu/economics/landsdown/cct/f2f/home/...2/8/2012 2 1. Be consistent 2. Not conflict with established fact or theory 3

2/8/2012

4

•Causal Mechanisms & Background Knowledge

li i hEg. Can eliminate the hypothesis that as I jump up and down at home can cause an earthquake, our background knowledge makes this unlikely.

Unless I had a big dinner 

•The Best Diagnosis Method

Eg. Sort of like clue, or twenty questions.

• Controlled Cause‐to‐Effect Experiments

• Look at a control group and give the other half the suspected “cause”

Diff i h

•Confirming Causal Hypotheses

• Difference in the frequency

• The difference between how much the effect occurred between the control and the “infected”

• Statistically Significant

• If at a certain confidence level, the d is higher, then it’s not just due to chance!

Page 5: PowerPointdeveloped bySarahLandsdown ©2011coursecontent.ntc.edu/economics/landsdown/cct/f2f/home/...2/8/2012 2 1. Be consistent 2. Not conflict with established fact or theory 3

2/8/2012

5

• An Example• Looking at weight loss drug, two groups of 1,500 are compiled: one is a control, the other is given the drug.

• The incidence that the control group lost weight was 25%.  The other group  had 26% of their population lose weight.  Is it statistically significant at 95% confidence level?

• What if the experimental group had weight loss at 42%?

•Controlled Cause‐to‐Effect

Number in Experimental Group (with similar sized control group)

Approximate Figure that d must exceed to be statistically significant (% points)

500 6

1,000 4

1,500 3

Other Methods

• Nonexperimental Cause‐to‐Effect Studies

• Use people who have already voluntarily taken the causal agent.

•Confirming Causal Hypotheses

• Nonexperimental Effect‐to‐Cause StudiesCause Studies

• Compare those that have the effect to those that don’t to see what the difference is.

• Experiments on Animals

1. Complicated

2. Not conflict with established fact or theory

3. Be testable

4. Not be circular

5. Cum hoc, ergo propter hoc

6. Post hoc, ergo propter hoc

f d l b b l d l

•Mistakes in Causal Reasoning

7. Confusing Conditional Probabilities in Medical Tests

8. Overlooking Statistical Regression

9. Proof by Absence of Disproof

10. Appeal to Anecdote

11. Confusing Explanations with Excuses

Page 6: PowerPointdeveloped bySarahLandsdown ©2011coursecontent.ntc.edu/economics/landsdown/cct/f2f/home/...2/8/2012 2 1. Be consistent 2. Not conflict with established fact or theory 3

2/8/2012

6

•George has tested positive for bladder cancer. The test is 90% accurate (90% of those who have bladder cancer test positive, and 10% of those who don’t have it, test positive (false positives)). How probable is it that George has bladder cancer?

A. Extremely probable

Let’s see!

B. Very probable

C. Probable

D. About 50/50

E. Improbable

16 © 2009 McGraw-

Hill Higher

Well, just because 90% of people with bladder cancer test positive, it does NOT follow that 90% of those who test positive have bladder cancer!

17 © 2009 McGraw-

Hill Higher

Suppose the rate for bladder cancer is 1%.  Then, for every 1000 people:

About 10 will have bladder cancer.

Who will probably test positive.

And 990 people will not have bladder cancer.

And 10% of those will test positive (99).

So: a total of 109 will test positive.

Of the 109 who test positive, 10 will have bladder cancer: That’s 9%

Given this information, the probability George has bladder cancer if he tests positive is really just 9%—not 90%.

18 © 2009 McGraw-Hill Higher Education. All rights

Page 7: PowerPointdeveloped bySarahLandsdown ©2011coursecontent.ntc.edu/economics/landsdown/cct/f2f/home/...2/8/2012 2 1. Be consistent 2. Not conflict with established fact or theory 3

2/8/2012

7

•# 1: OVERLOOKING REGRESSION

What’s regression?

“Regression” refers 

19 © 2009 McGraw-Hill Higher Education. All rights

to the fact that:

“Statistically, things tends to return to normal.”

•For example, the sons of tall fathers are, on average, shorter than their fathers.

And the sons of short fathers on average are taller than their fathers.

In both cases, the sons’ average height will be closer than their fathers’ to the mean.

20 © 2009 McGraw-

Hill Higher

•And suppose we test this class, and find your average pulse rate is higher than normal. If we then test you again, your average will be lower closer to the humanbe lower, closer to the human mean.

On the other hand, if it was lower than normal on the first test, it will go up on the second test.

21 © 2009 McGraw-

Hill Higher

Page 8: PowerPointdeveloped bySarahLandsdown ©2011coursecontent.ntc.edu/economics/landsdown/cct/f2f/home/...2/8/2012 2 1. Be consistent 2. Not conflict with established fact or theory 3

2/8/2012

8

Things tend toreturn to normalreturn to normal

22© 2009 McGraw-Hill Higher Education. All rights

•Proof by Absence of Disproof

Eg. Because you can’t prove that Big Foot doesn’t exist doesn’t give me more proof that he does.

•Appeal to Anecdote

b hEg. Just because I have observed that I haven’t gotten a sinus infection once I started using a neti pot, doesn’t mean that they are a cure for sinus infections.

Page 9: PowerPointdeveloped bySarahLandsdown ©2011coursecontent.ntc.edu/economics/landsdown/cct/f2f/home/...2/8/2012 2 1. Be consistent 2. Not conflict with established fact or theory 3

2/8/2012

9

•Confusing Explanations with Excuses

Eg. Someone speaking on the Columbine shooting as a case against bullying.