CRITICAL

THINKING
COURSE MANUAL
2023 Fall

This document is still being written and revised.

• If you are the first to report a minor typo (a spelling/grammar/numbering mistake, a
formatting error, etc.), you will get a 0.5 mark bonus in the course.
• If you are the first to report a major typo or error, you will get a 1 mark bonus in the course.
• If you suggest a way to make a sentence or diagram more clear and I use your suggestion,
you will get a 0.5 mark bonus in the course.

Copyright © 2023 Lyle Crawford

-1-
 -CONTENTS-

INTRODUCTION 3

1. ARGUMENTS 4

2. CATEGORY LOGIC 16

3. STATEMENT LOGIC 53

4. EXPLANATION AND CONFIRMATION 81

5. CONFIRMATION BIAS 111

6. PROBABILITY 122

7. BAYES’ RULE 149

8. ANALOGIES 186

9. GENERALIZING FROM A SAMPLE 202

10. CAUSE AND EFFECT 220

-2-
 -INTRODUCTION-

This is a course about how to think, and how to understand and evaluate what other people think. It
won’t tell you what to think. You won’t prepare for exams by memorizing a lot of facts. Instead, you’ll
develop general thinking skills for deciding what to believe and what to do. These are skills for you to
use in any situation where having true beliefs matters, either because you simply want to know what’s
true or because actions guided by true beliefs are the way best way to achieve your goals. These are
also skills for clear and effective communication.

Here’s an overview of what we’ll do in this course:

Arguments
An argument presents reasons to hold a belief. It’s an opportunity to learn from someone else, by
seeing that their belief is well-founded. But it’s also an opportunity to discover that someone’s reasons
do not support their belief. To take advantage of arguments while minimizing the risk of being misled
by them, we need to analyze and evaluate them. To use arguments to show others that we have our
own good reasons for a belief, we need skills for constructing and expressing them.
Logic and Probability
Logic and probability include some of the most basic rules of good reasoning. They help us notice
when our beliefs should lead us to a new belief and when they conflict with one another. And they
help us to see how strongly we should hold a belief based on the strength of our other beliefs, which
is crucial for making good decisions and for thinking about risk and reward.
Explanation and Confirmation
Explanation and confirmation are fundamental to scientific thinking, to the careful use of evidence to
arrive at reasonable beliefs about the world. There are standard patterns of explanatory and
confirmatory reasoning that we can learn to use and recognize. And there are common forms of bias
that are difficult to suppress by simple effort but that we can learn to notice and counteract.
Samples and Experiments
We rely on samples to learn about the world from the limited data that’s often available to us, and
we’re confronted with their use in the statistics we see every day. They’re also used in the experiments
by which we discover causal relationships and work out how to control the world to our advantage,
such as in medical and environmental science.

-3-
 ~1~
-ARGUMENTS-

Several units in this course develop the critical thinking skills of reading, writing, and evaluating
arguments, so a general overview of arguments is a good place to begin.

i. -STATEMENTS-

To understand what an argument is, we need the more basic concept of a statement.

Three Types of Sentences

Here are two types of sentences: question (asking for information); command (telling someone to
do something).

Question 1.1: Why did the Maya civilization decline?

Command 1.2: Walk backwards with your eyes closed.

A third type of sentence is a statement. It asserts (says) that something is a fact.

Statement 1.3: There are twelve continents.

Statement 1.4: Airplane business class seating is nicer than economy seating.

Every statement has a truth value: either true or false. This makes statements different from
questions and commands, which are neither true nor false. Compare these sentences:

It’s true that why did the Maya civilization decline. [It’s true that QUESTION.]
It’s true that walk backwards with your eyes closed. [It’s true that COMMAND.]
It’s true that there are twelve continents. [It’s true that STATEMENT.]

Only the last sentence makes sense because only the statement has a truth value (false).

-4-
ii. -ARGUMENTS-

We often hear that two people “had an argument” or that

someone “won an argument”. In ordinary language, an argument
is a dispute or disagreement. However for us, an argument is not
a dispute, and it’s not something that anyone wins or loses.

An argument is a set of statements: one or more premise statements given as support for (reasons
to believe) a conclusion statement. The conclusion may be anywhere in the argument. The conclusion
is underlined in Arguments 1.5-1.6.

Argument 1.5: Flying in an airplane is more dangerous than riding in a car since
airplanes fly 12 km up in the sky whereas cars stay on the ground.

Argument 1.6: The death penalty is very cruel and does not
prevent crimes, so it should be abolished.

To make an inference is to reason from premises to a conclusion. There are many words and phrases
– inference indicators – that help us notice an inference. Argument 1.5 says “[Conclusion] since
[Premise]”; Argument 1.6 says “[Premises] so [Conclusion]”. There are many others, such as because,
consequently, therefore, and for that reason. And some arguments have no inference indicator.

Argument 1.7: Social networks (such as Facebook) are where most citizens get their
news now. They should be controlled by laws ensuring the accuracy of news stories.

The premise “Social networks are where most citizens…” supports the conclusion “They should be
controlled…”.

-5-
iii. -RECONSTRUCTING ARGUMENTS-

Standard Form

Writing an argument in standard form puts its premises above a line and its conclusion below. We
can reconstruct an argument by reading (or hearing) it carefully and putting it in standard form.

Argument 1.8: The price of gold will go up soon, since

investors are getting nervous about the market. They usually
sell stocks and buy gold when they get nervous.

Argument 1.8 (standard form):

• Investors are getting nervous about the market.
• Investors usually sell stocks and buy gold when they get nervous about the market.

The price of gold will go up soon.

Argument 1.9 is the same argument as Argument 1.8 but includes some extra information. These
statements may introduce or elaborate on statements in the argument. They are neither premises nor
conclusions and should be left out of the standard form reconstruction.

Argument 1.9: Even though the price of gold is already nearly as high as it has ever
been, it will go up soon, since investors are getting nervous about the stock market. They
usually sell stocks and buy gold when they get nervous. People seem to believe that just
because gold is shiny, it is a safe investment.

-6-
 Implicit Statements

People often do not say everything they think when giving an argument. A statement that someone
thinks of as part of their argument, but does not write or speak, is implicit.

Argument 1.10: We can’t save him now. He hasn’t had

a heartbeat for 5 minutes.

There is a premise in Argument 1.10 that is not stated. Its exact wording and meaning are not
completely clear, but the general idea is clear enough. It must be something like:

No one who hasn’t had a heartbeat for 5 minutes can be saved.

If someone hasn’t had a heartbeat for 5 minutes then they cannot be saved.
Very few people who haven’t had a heartbeat for 5 minutes can be saved.
If someone’s had no heartbeat for 5 minutes then it’s very unlikely they can be saved.

Some such statement must be part of the speaker’s meaning in Argument 1.10. It’s implicit.

Extended Arguments

An extended argument has a main argument with a final conclusion, and at least one sub-
argument. The sub-argument has an intermediate conclusion, which is a premise in the main
argument. Starting in Unit 4, to make reconstructions efficient, we’ll use statement numbers (①, ②,
etc.) for premises, conclusions, and extra information.

Argument 1.11: ① We should buy gold. ② It’s a good investment, since ③ everything
shiny is a good investment, and ④ gold is very shiny. And ⑤ we should buy it if it’s a
good investment.

Main Argument Sub-argument

② ③
⑤ ④

① ②

-7-
iv. -EVALUATING ARGUMENTS-

To evaluate an argument is to judge its value, to say whether it is good or bad. We ask two questions:
1) Are the premises true? 2) Do the premises support the conclusion? With a good argument, we can
answer Yes to both questions. We’ll do this a bit in Units 2-3 and more in Units 8-10.

Are the premises true?

Argument 1.12: All paintings are by

Picasso, and Guernica (1937) is a painting.
Hence Guernica is by Picasso.

(Image: Wikimedia)

The premises of Argument 1.12 support its conclusion. But its first premise is not true: not all paintings
are by Picasso. So it is a bad argument (even though the conclusion is true).

Do the premises support the conclusion?

Premises support the conclusion when: If the premises are true, they are a reason to believe the
conclusion.

Argument 1.13: All Winter Olympics have skiing events, and

the 2018 Olympics in PyeongChang, Korea, had skiing events.
Therefore the 2018 Olympics was a Winter Olympics.

The premises of Argument 1.13 are true. But its premises do not support its conclusion. So it’s a bad
argument (even though the conclusion is true).

Perhaps you aren’t sure that Argument 1.13 is bad in this way. Or perhaps you are sure but you’re
not sure how to show or explain it. That’s fine. Unit 2 covers this kind of argument. There we’ll learn
some tools for analyzing and evaluating arguments like this one.

-8-
 Do we know any relevant background information?

Knowing that an argument has true premises that support its conclusion does not automatically make it
rational to believe that conclusion. This is because support does not always mean guarantee. In some cases,
it does. We’ll look at arguments like that (deductive) in Units 2-3. In other cases, it does not. We’ll look at
those other types of arguments in Units 4 and 8-10. With these arguments, it can be very important to
consider background information in addition to the premises.

-9-
 -UNIT 1 SKILLS-

You must be able to:

• Recognize and write examples of questions, commands, and statements.

• Identify inference indicators, conclusions, and premises in an argument.

• Reconstruct an argument in standard form.

- 10 -
 -QUICK TEST QUESTIONS-

1) The conclusion of an argument…

a) …briefly summarizes the premises. c) …supports one or more premises.
b) …is supported by one or more premises. d) …comes at the end of an argument.

2) How is an argument written in standard form?

a) It uses a standard inference indicator c) It’s written as a paragraph with the
such as because. conclusion stated clearly.
b) It’s written in a form that meets a very d) The premises are written above a line
high standard. with the conclusion below it.

3) An implicit statement is…

a) …part of the argument and is written. c) …not part of the argument and is not
b) …not part of the argument but is written.
written. d) …part of the argument but is not written.

4) An extended argument always contains…

a) …an intermediate conclusion and a final c) …an implicit premise and an intermediate
conclusion. conclusion.
b) …an intermediate conclusion and a final d) …an intermediate premise and a final
premise. premise.

5) To evaluate an argument is to judge whether…

a) …its conclusion is true or false. c) …the argument is good or bad.
b) …the argument is true or false. d) …its conclusion is good or bad.

6) Which things are true or false?

a) Premises and arguments; not conclusions. c) Premises, conclusions, and arguments.
b) Premises and conclusions; not arguments. d) Arguments and conclusions; not premises.

7) What are the features of a good argument?

a) The premises are true; the conclusion is c) The premises are true; the premises
true. support the conclusion.
b) The argument is true; the premises d) The conclusion is true; the conclusion
support the conclusion. supports the premises.

- 11 -
 -PRACTICE QUESTIONS-

1) Is the sentence a question, a command, or a statement?

a) Every person who’s gone to the Moon was a woman.
b) Please fly me to the Moon.
c) We should send humans to Mars instead of the Moon.
d) Could humans live on Mars without getting bored?
e) Why does the Moon have craters on it?
f) Humans first landed on the Moon in 1785.
g) Stop spending money on stupid things like sending humans to the Moon.
h) The volcano near our city might explode this year.
i) You are allowed to use your phone during the exam.
j) Why does our instructor always have a headache?
k) Quickly hand me the screwdriver that fits these screws.
l) The deepest part of the ocean is in English Bay, Vancouver.
m) Just get rid of that gross leftover take-out food in the refrigerator.
n) How does a representative democracy like Canada work?
o) The end of the Bronze Age may have been caused by climate change.

2) Identify the conclusion and inference indicators if there are any.

a) College courses could be taught by robots instead of humans. Robots can click PowerPoint
slides and mark multiple choice exams just as well as humans can.
b) Since cheetahs are beautiful wild animals that belong in their home in Africa, it would be good
to have better enforcement of the laws that prohibit capturing them for pets.
c) It would be better for people to use transit instead of driving. Just consider the fact that, with
millions of cars on the road, no one can get where they’re going anyway.
d) Obviously K2 is a more challenging climb than Mount Everest. Serious mountain climbers
admire people who climb K2 more than they admire people who climb Mount Everest.
e) The Parthenon Marbles in the British Museum were taken from Greece in the 19 th century,
and now Greece wants them back, so Britain has no right to keep them.
f) The IUD (Intrauterine Device) is the most reliable form of contraception [birth control]. As
well as having a low failure rate, IUDs don’t depend on women taking a pill every day.
g) In our society there are millions of “soft” (unprotected) targets for terrorists who are willing
to die in an attack; consequently we cannot stop terrorism without restricting people’s
freedoms.
h) In 2009, the cable news channels reported a crazy, false story about a little boy who floated
away in a homemade balloon. The people at the cable news channels are idiots.

- 12 -
i) Earth is only about 10,000 years old. The Bible says so, and the Bible was written by the
creator of the universe.
j) If the new guy is working today, some customer will get their order screwed up. And a customer
did get their order screwed up. So the new guy is working today.
k) Taxes are really just a kind of theft: the government robs people of their money and calls it a
tax so that it doesn’t sound so bad. That’s why taxation is wrong.
l) Sending humans to explore Mars would be far more expensive than sending robots. Unlike
robots, humans need protection from the cold vacuum [lack of air] and dangerous radiation of
outer space, and a spaceship that can do that would cost a lot.
m) Buddy’s Pizza and Bobby’s Pizza are both cheap, greasy pizza restaurants, and Buddy’s pizzas
are gross. Therefore Bobby’s pizzas will be gross, too.
n) Sending humans to explore other planets would be very expensive, and most of that money
would be spent on machines and equipment whose only purpose is to keep the humans alive.
So we should send robots to explore outer space instead, as they can take photographs and
do science experiments without the air, food, and warmth that humans need.
o) Since the American government is evil and wanted an excuse to invade Iraq for its oil, probably
it planned the 2001 World Trade Center attack and made it look as though terrorists did it.
p) Canada’s economy will be affected by a drop in oil prices due to the fact that Canada exports
oil from the tar sands of Alberta, and it is so expensive to extract oil from tar sands that it is
profitable only if the price of oil is high.
q) Even though they come to the surface for air to breathe, whales have fins and tails for
swimming and live their whole lives in the ocean, and anything that swims and lives in the
ocean is a fish. It follows that whales are fish.
r) Those herbs that my friend gave me must have cured my arthritis [joint pain]. The proof is
that my arthritis went away shortly after I started eating them every day.

- 13 -
 -ANSWER KEY-

QUICK TEST
1) b 2) d 3) d 4) a 5) c 6) b 7) c

1)
a) Statement f) Statement k) Command
b) Command g) Command l) Statement
c) Statement h) Statement m) Command
d) Question i) Statement n) Question
e) Question j) Question o) Statement

2) Conclusion is underlined; inference indicator is bold.

a) College courses could be taught by robots instead of humans. Robots can click PowerPoint
slides and mark multiple choice exams just as well as humans can.
b) Since cheetahs are beautiful wild animals that belong in their home in Africa, it would be
good to have better enforcement of the laws that prohibit capturing them for pets.
c) It would be better for people to use transit instead of driving. Just consider the fact that,
with millions of cars on the roads, no one can get where they’re going anyway.
d) Obviously K2 is a more challenging climb than Mount Everest. Serious mountain climbers
admire people who climb K2 more than they admire people who climb Mount Everest.
e) The Parthenon Marbles in the British Museum were taken from Greece in the 19 th century,
and now Greece wants them back, so Britain has no right to keep them.
f) The IUD (Intrauterine Device) is the most reliable form of contraception [birth control]. As
well as having a low failure rate, IUDs don’t depend on women taking a pill every day.
g) In our society there are millions of “soft” (unprotected) targets for terrorists who are willing
to die in an attack; consequently we cannot stop terrorism without restricting people’s
freedoms.
h) In 2009 the cable news channels reported a crazy false story about a little boy who floated
away in a homemade balloon. The people at the cable news channels are idiots.
i) Earth is only about 10,000 years old. The Bible says so, and the Bible was written by the
creator of the universe.
j) If the new guy is working today, some customer will get their order screwed up. And a customer
did get their order screwed up. So the new guy is working today.
k) Taxes are really just a kind of theft: the government robs people of their money and calls it a
tax so that it doesn’t sound so bad. That’s why taxation is wrong.

- 14 -
l) Sending humans to explore Mars would be far more expensive than sending robots. Unlike
robots, humans need protection from the cold vacuum [lack of air] and dangerous radiation of
outer space, and a spaceship that can do that would cost a lot.
m) Buddy’s Pizza and Bobby’s Pizza are both cheap, greasy pizza restaurants. And Buddy’s pizzas
are gross. Therefore Bobby’s pizzas will be gross, too.
n) Sending humans to explore other planets would be very expensive, and most of that money
would be spent on machines and equipment whose only purpose is to keep the humans alive.
So we should send robots to explore outer space instead, as they can take photographs and
do science experiments without the air, food, and warmth that humans need.
o) Since the American government is evil and wanted an excuse to invade Iraq for its oil, probably
it planned the 2001 World Trade Center attack and made it look as though terrorists did it.
p) Canada’s economy will be affected by a drop in oil prices due to the fact that Canada exports
oil from the tar sands of Alberta, and it is so expensive to extract oil from tar sands that it is
profitable only if the price of oil is high.
q) Even though they come to the surface for air to breathe, whales have fins and tails for
swimming, and they live their whole lives in the ocean, and anything that swims and lives in
the ocean is a fish. It follows that whales are fish.
r) Those herbs that my friend gave me must have cured my arthritis [joint pain]. The proof is
that my arthritis went away shortly after I started eating them every day.

- 15 -
 ~2~
-CATEGORY LOGIC-

i. -UNIVERSAL GENERALIZATIONS-

Categories and Individuals

A category is a type of thing. We’ll picture it as a container, a circle. To show that an individual
thing is a member of the category, we’ll show the individual as an X inside a labeled circle.

Statement 2.1: The 1917 Halifax Explosion (X) was a disaster. Disasters

Statement 2.2: Wednesday (X) is not a month. Months

(Wednesday (X) is a non-month.) X

The X means “something exists here”. A category may have no members at all – it may be empty.
For example, the category Canadian $3 Bills is empty because there no Canadian $3 bills.

Universal Generalizations

A universal generalization relates two categories to each other and says: “not even one exception to
this”. It’s written with a universal quantifier All or No (or an alternative word such as Every or None).

All A are B. No A are B.

Here A and B stand for the subject category and the predicate category.


These are called Euler diagrams. This unit presents a simplified version of the category logic taught in a formal logic course.

- 16 -
A statement has a subject (what the statement is about) and a predicate (what the subject does or is).
Normally to find the predicate, we find the verb. Here are some examples with the predicate underlined:

The 1917 Halifax explosion was a disaster. All bacteria cause disease.
Bob loses his temper very easily. No cities span two continents.
The economy of Azmakia grew by 5% last year. Only birds fly.

The examples on the right are universal generalizations. The quantifiers tell us how two categories are
related to one another. We can use a diagram with two labeled circles to show this relationship.

Statement 2.3: All bacteria cause disease. Disease-

All Bacteria are Disease-Causers. Bacteria Causers

The All quantifier tells us that the subject category Bacteria is completely contained within the
predicate category Things that Cause Disease (we’ll use a short form: Disease-Causers).

Statement 2.4: No cities span two continents. Places that

Span Two
No Cities are Places that Span Two Continents.
Continents

Cities

The No quantifier tells us that the subject category Cities is completely separate from the predicate
category Places that Span Two Continents.

Statement 2.5: Only birds fly.

Birds
Only Birds are Flying Animals. Flying
Animals

Only functions as a universal quantifier. Like All, Only describes a containment relationship, but Only
does not mean All. Here Only tells us that the subject category Birds completely surrounds the
predicate category Flying Animals.

- 17 -
 Logical Equivalence

When two categorical statements are logically equivalent, they have exactly the same diagram, and
they are either both true or both false.

There are two kinds of diagrams that show a universal relationship between two categories: complete
containment, or complete separation. For each diagram there is a pair of logically equivalent
statements using each quantifier. For most people, the All and No statements are easiest to see in the
diagram, and the Only statements takes a bit more practice.

Logically Equivalent Universal Generalizations

All A are B. No A are B.

A
All Non-B are Non-A. No B are A.
A
Only B are A. All A are Non-B.

Only Non-A are Non-B. All B are Non-A.

B Only Non-B are A. B
No A are Non-B.
No Non-B are A. Only Non-A are B.

Statement 2.3: All Bacteria are Disease-Causers.

Only Disease-Causers are Bacteria.

No Bacteria are Non-(Disease-Causers). Bacteria

The categories switch positions in the logically equivalent All and

Only statements.
Disease-Causers

All Non-(Disease-Causers) are Non-Bacteria.

Only Non-Bacteria are Non-(Disease-Causers).

No Non-(Disease-Causers) are Bacteria.

- 18 -
 Statement 2.4: No Cities are Places that Span Two Continents.

Cities
No Places that Span Two Continents are Cities.

All Cities are Non-(Places that Span Two Continents).

Places that
All Places that Span Two Continents are Non-Cities. Span Two
Continents

Only Non-(Places that Span Two Continents) are Cities.

Only Non-Cities are Places that Span Two Continents.

An All statement can also be combined with a different (not logically equivalent) Only statement. The
combined statements have a combined diagram.

Disease-
Bacteria
Statement 2.3: All bacteria cause disease. Causers

+ +

Statement 2.6: Only bacteria cause disease. Bacteria Disease-

Causers

= =

Statement 2.7: All and only bacteria cause disease.

Disease-
Bacteria
Causers

All and only… says that the two categories perfectly overlap. They contain exactly the same things.

- 19 -
ii. -DEDUCTIVE ARGUMENTS-

An argument consists of statements (premises) given in support of another statement (the

conclusion). In a good argument, the premises are true and they support the conclusion. If someone
believes the premises, they have a reason to believe the conclusion, as well.

• Premise
• Premise

Conclusion

Deductive arguments are arguments that are supposed to be valid. The premises of a valid
argument support the conclusion absolutely. The premises guarantee the conclusion. A valid argument
is "truth-preserving”: If its premises are true, its conclusion is definitely true, as well.

Logic is concerned with valid argument patterns, or argument structures. Here is an example:

• All A are B.
B
• X is an A. A
X
X is a B.

This argument is written just as a pattern. The A and B categories, and the individual X, could be
anything. However even without knowing what A, B, and X are, we can see that the argument is
truth-preserving. A diagram showing its premises true automatically shows its conclusion true, as
well. Any argument that follows this pattern, regardless of what A, B, and X are, is valid.

Short deductive arguments with patterns like this one are often called syllogisms. In the next section
we’ll look at some different syllogisms that have a universal generalization premise.


Vocabulary: In ordinary language, valid has other meanings: true, relevant, important, etc. A scientific test is valid if it
measures what it’s supposed to measure. Deductive validity is different from all of these.

- 20 -
iii. -CATEGORICAL SYLLOGISMS-

Argument Patterns with All

Here is a valid pattern and a similar but invalid pattern (logical fallacy), with examples below.

• All A are B. • All A are B.

• X is an A. • X is a B.

X is a B. X is an A.

VALID INVALID

Argument 2.8 (VALID)

• All disasters are earthquakes. Earthquakes
Disasters
• The Halifax Explosion (X) was a disaster.
X
The Halifax Explosion (X) was an earthquake.

The diagram shows why Argument 2.8 is valid: If the premises are true, the conclusion
must be true (Halifax Explosion (X) inside Earthquakes).

This argument is valid even though its conclusion is false. A valid argument may have a
false conclusion if at least one of its premises is false. Here the first premise is also false.

Argument 2.9 (INVALID)

Food- Food-
• Every restaurant serves food.
Restau Servers Restau Servers
• McDonald’s (X) serves food. rants rants
X X
McDonald’s (X) is a restaurant.
Conclusion True Conclusion False!

The diagrams show why Argument 2.9 is invalid: If the premises are true, the conclusion
might be true (left diagram), but it is also might be false (right diagram) – the truth of the
premises does not guarantee the truth of the conclusion.

- 21 -
Here is another valid pattern with All and a similar but invalid pattern, with examples below.

• All A are B. • All A are B.

• X is not an A. • X is not a B.

X is not a B. X is not an A.

INVALID VALID

Argument 2.10 (VALID)

• All bacteria cause disease. Disease-
Bacteria Causers
• Yersinia pestis (X) does not cause disease. X

Yersinia pestis (X) is not a bacterium.

The diagram shows why Argument 2.10 is valid: If the premises are true (Yersinia
pestis (X) outside Disease-Causers), the conclusion must be true (X outside Bacteria).

Argument 2.11 (INVALID):

Superhero Superhero
• All great films depict superheroes. Depictions Depictions
Great Great
• Elf (X) is not a great film. Films Films
X
Elf (X) does not depict superheroes. X
Conclusion False! Conclusion True

The diagrams show why Argument 2.11 is invalid: If the premises are true, the
conclusion might be true (right diagram), but it is also might be false (left diagram) –
the truth of the premises does not guarantee the truth of the conclusion.

- 22 -
 Argument Patterns with No

With No, again there are two valid patterns and two similar but invalid patterns.

• No A are B. • No A are B. • No A are B. • No A are B.

• X is an A. • X is a B. • X is not an A. • X is not a B.

X is not a B. X is not an A. X is a B. X is an A.

VALID VALID INVALID INVALID

Argument 2.12 (VALID)

• No medieval inventions changed the world. Medieval World-
Inventions Changing
• The mechanical clock (X) changed the world. Inventions

The mechanical clock (X) wasn’t a medieval X

invention.

The diagram shows why Argument 2.12 is valid: If the premises are true (the
mechanical clock (X) inside World-Changing Inventions), the conclusion must be true
(X outside Medieval Inventions).

Argument 2.13 (INVALID) Things Things

• Nothing worth doing is easy. X Worth Worth
Doing Doing
• Juggling chainsaws (X) is not easy.
Easy Easy
X Things
X Things
Juggling chainsaws (X) is worth doing. X

Conclusion True Conclusion False!

The diagrams show why Argument 2.13 is invalid: If the premises are true, the
conclusion might be true (left diagram), but it also might be false (right diagram) – the
truth of the premises does not guarantee the truth of the conclusion.

- 23 -
iv. -COUNTEREXAMPLES-

A universal generalization is false if there is even one exception to it. (A true universal generalization
has no exceptions.) An exception to a generalization is called a counterexample.

With two categories A and B, everything in the world is one of four types of things:

A B A Non-B

Non-A B Non-A Non-B

In the diagram for a universal generalization, there are separate regions for only three types of things.
The fourth type of thing, which does not fit into the diagram, is the counterexample.

In the diagram for “All A are B”, there is no region for [A, Non-B] (inside A circle, outside B circle).
So that’s the counterexample.

A B A Non-B
B
A
Non-A B Non-A Non-B

In the diagram for “No A are B”, there is no region for [A, B] (inside both circles). So that’s the
counterexample.

A B A Non-B
A B
Non-A B Non-A Non-B

- 24 -
 Case #1: “Only birds fly.”

Statement 2.5: Only birds are flying animals.

Birds Flying
Animals

Statement 2.5 is false. A bat (a flying animal

that is not a bird) is a counterexample because
a bat (X) does not fit into the diagram. The
diagram must be corrected to show a bat. The
Flying Animals circle must extend outside of the
Birds circle.

Bird Flying Bird Non-(Flying Correction Diagram

Animal Animal)

Birds
Flying
Non-Bird Flying Non-Bird Non-(Flying
Animals
Animal Animal)
X

A bat proves that:

• Some flying animals are not birds.
• Not only birds fly.

- 25 -
 What about an owl (a flying bird)? No, an owl
(X) is not a counterexample because it fits into
the diagram.

Bird Flying Bird Non-(Flying

Animal Animal)

Birds Flying
Animals

Non-Bird Flying Non-Bird Non-(Flying X

Animal Animal)

What about a penguin (a bird that doesn’t fly)?

No, a penguin (X) is not a counterexample
because it fits into the diagram.

Bird Flying Bird Non-(Flying

Animal Animal)

Birds Flying
Animals

Non-Bird Flying Non-Bird Non-(Flying

Animal Animal) X

- 26 -
 What about a rock (a non-bird that doesn’t
fly)? No, a rock (X) is not a counterexample
because it fits into the diagram.

Bird Flying Bird Non-(Flying

Animal Animal)

Birds Flying
Animals

Non-Bird Flying Non-Bird Non-(Flying X

Animal Animal)

What about a dragon (a flying animal that’s not a

bird)? A dragon would be a counterexample
because it would not fit into the diagram. But
dragons are mythical – they don’t exist! A dragon
doesn’t disprove Statement 2.5.

Bird Flying Bird Non-(Flying

Animal Animal)

Birds Flying
Animals

Non-Bird Flying Non-Bird Non-(Flying

Animal Animal)

- 27 -
 Case #2: “No cities span two continents.”

Statement 2.4: No cities span two continents. Places that

Cities
Span Two
Continents

Statement 2.4 is false. Istanbul (a city that spans

two continents [Europe and Asia]) is a
counterexample because Istanbul (X) does not fit
into the diagram. The diagram must be corrected
to show Istanbul. The two circles must overlap.
(Image: Freepik.com)

City Place City Non-(Place

that spans… that Spans…) Correction Diagram

Places that
Non-City Place Non-City Non-(Place Cities Span Two
That Spans… that Spans…) X Continents

Istanbul proves that:

• Some cities span two continents.
• Not no cities span two continents.*

*
“Not no…” is not a standard accepted construction in English.

- 28 -
 Case #3: “All and only bacteria cause disease.”

Statement 2.7: All and only bacteria cause disease.

Disease-
Bacteria
Causers

Statement 2.7 is doubly false. There are

counterexamples of two types: 1) soil bacteria
(bacteria that don’t cause disease), and 2) UV
radiation from the Sun (non-bacteria that
causes disease). Soil bacteria (X1) and UV
(Image: drmicrobe)
radiation (X2) don’t fit into the statement’s
diagram. It must be corrected. The two circles
must pull apart from each other.

Bacteria Disease- Bacteria Non-

Causer (Disease- Correction Diagram
Causer)

Bacteria Disease-
Non- Disease- Non- Non- Causers
Bacteria Causer Bacteria (Disease-
Causer) X1 X2

Soil bacteria proves that:

• Some bacteria do not cause disease.
• Not all bacteria cause disease.

UV radiation from the Sun proves that:

• Some disease-causers are not bacteria.
• Not only bacteria cause disease.

- 29 -
 -UNIT 2 SKILLS-

You must be able to:

• Write the two categories in a universal generalization and diagram it.

• Write logically equivalent statements for a universal generalization.

• Construct two syllogisms (valid deductive arguments) from a universal generalization premise.

• Use a diagram (two circles and an X) to prove the invalidity of an argument.

• Think of counterexamples.

• Write a “Some…” statement for a false universal generalization and draw a diagram showing the

categories and counterexample.

- 30 -
 -QUICK TEST QUESTIONS-

1) Which is not a universal quantifier?

a) No c) Only
b) All d) Some

2) Which statement is logically equivalent to “Only A are B”?

a) Only B are A. c) All A are B.
b) No B are A. d) All B are A.

3) Which diagram shows that all and only A are B?

a) A B c) A B

b) A B d) A B

4) Which is not logically equivalent to “All A are B”?

a) Only A are B. c) No Non-B are A.
b) No A are Non-B. d) All Non-B are Non-A.

5) Describe a counterexample to “Only A are B”.

a) A B c) Non-A B
b) Non-A Non-B d) A Non-B

6) A true universal generalization has how many counterexamples?

a) Not enough info. c) 1
b) ≥1 d) 0

7) Bob thinks that “All A are B”. Abby disagrees: she thinks that X is a counterexample. What
could Bob say to Abby?
a) “You’re wrong, X is a B.” c) “You’re wrong, X is an A.”
b) “You’re wrong, X is not a B.” d) None of these.

8) Which is impossible with a valid argument?

a) True premises; false conclusion. c) False premises; false conclusion.
b) False premises; true conclusion. d) True premises; true conclusion.

- 31 -
 -PRACTICE EXAMS-

PRACTICE EXAM 1

1) Complete two valid syllogisms with different argument patterns. For each argument, use the given
universal generalization as Premise 1 and complete Premise 2 and the Conclusion using the given
subject. Below the arguments, draw a single diagram showing the premises of both arguments
true (two circles, and X1 and X2).

• [Premise 1] Every legitimate election offers voters mail-in ballots.

• [Premise 2] Azmakia’s election (X1)… • [Premise 2] Pseudorica’s election (X2)…

_______________________________ _______________________________

_______________________________ _______________________________

[Conclusion] ________________________ [Conclusion] ________________________

__________________________________ __________________________________

Diagram

2) Draw a labeled diagram that proves the given argument is invalid and give a brief explanation of
how it proves this.

• Only coffee shops that follow public Diagram

health rules stay open.
• Constant Coffee follows public health
rules.

Constant Coffee stays open.

- 32 -
3) For the given universal generalization statement, write the two categories used in the statement
and write logically equivalent statements using the given quantifiers.
Once you have assigned letters (A, B) to the categories, you may use either the full labels or just
the letter labels for the rest of your answer.

No industries extract natural resources from the ground.

Categories

A: ________________________________________________________________________

B: ________________________________________________________________________

Logically Equivalent Statements

All…_______________________________________________________________________

Only…_____________________________________________________________________

Give a counterexample to the statement above and say what it proves. Draw a diagram of the
original generalization. Draw a corrected diagram that includes your counterexample (X).

Counterexample (X): _________________________________________________________

X proves that: Some…________________________________________________________

Diagram of Original Generalization Diagram of Correction (including X)

- 33 -
 PRACTICE EXAM 2

1) Complete two valid syllogisms with different argument patterns. For each argument, use the given
universal generalization as Premise 1 and complete Premise 2 and the Conclusion using the given
subject. Below the arguments, draw a single diagram showing the premises of both arguments
true (two circles, and X1 and X2).

• [Premise 1] Only Biosafety Level 4 medical labs can safely study samples of Ebola virus.

• [Premise 2] The Gotham Bioresearch • [Premise 2] Azmakia’s National

Institute’s lab (X1)… Virology Lab (X2)…
_______________________________ _______________________________

_______________________________ _______________________________

[Conclusion] ________________________ [Conclusion] ________________________

__________________________________ __________________________________

__________________________________ __________________________________

Diagram

2) Prove that the given argument is invalid using a labeled diagram with a brief explanation of how
it proves this.
• None of the icebergs that break off from Diagram
an Antarctic ice shelf float into tropical
waters.
• Iceberg B24c didn’t float into tropical
waters.

Iceberg B24c broke off from an Antarctic ice

shelf.

- 34 -
3) Write the two categories from the statement. Name two counterexamples with different
descriptions. Draw a diagram of the original generalization. Then draw a corrected diagram that
includes your counterexamples (X1 and X2). Once you have assigned letters (A, B) to the
categories, you may use either the full labels or just the letter labels for the rest of your answer.

All and only museums display artworks.

Categories

A: ________________________________________________________________________

B: ________________________________________________________________________

Counterexample (X1): _______________________________________________________

X1 proves that: Some…_______________________________________________________

Not all | only museums display artworks.

Counterexample (X2): ________________________________________________________

X2 proves that: Some…_______________________________________________________

Not all | only museums display artworks.

Diagram of Original Generalization Diagram of Correction (including X1 and X2)

- 35 -
 PRACTICE EXAM 3

1) Complete two valid syllogisms with different argument patterns. For each argument, use the given
universal generalization as Premise 1 and complete Premise 2 and the Conclusion using the given
subject. Below the arguments, draw a single diagram showing the premises of both arguments
true (two circles, and X1 and X2).

• [Premise 1] No companies that are good to invest in have a high-school drop-out CEO.

• [Premise 2] Wackadoodle Industries • [Premise 2] Flippers Spatula Co. (X2)…

(X1)…__________________________ _______________________________

_______________________________ _______________________________

_______________________________ _______________________________

[Conclusion] ________________________ [Conclusion] ________________________

__________________________________ __________________________________

__________________________________ __________________________________

Diagram

2) Prove that the given argument is invalid using a labeled diagram with a brief explanation of how
it proves this.

• Only tree-planting programs can Diagram

prevent climate change.
• The UN’s Trillion Tree Initiative is a
tree-planting program.

The UN’s Trillion Tree Initiative can prevent

climate change.

- 36 -
3) For the given universal generalization statement, write the two categories used in the statement
and convert the statement into logically equivalent statements using the given quantifiers.
Once you have assigned letters (A, B) to the categories, you may use either the full labels or just
the letter labels for the rest of your answer.

Every wildfire starts from a lightning strike.

Categories

A: ________________________________________________________________________

B: ________________________________________________________________________

Logically Equivalent Statements

Only…_____________________________________________________________________

No…_______________________________________________________________________

Give a counterexample to the statement above. Describe it (say what makes it a

counterexample). Draw a diagram of the original generalization. Then draw a corrected
diagram that includes your counterexample (X).

Counterexample (X): _________________________________________________________

X proves that: Some…________________________________________________________

Diagram of Original Generalization Diagram of Correction (including X)

- 37 -
 PRACTICE EXAM 4

1) Complete two valid syllogisms with different argument patterns. For each argument, use the given
universal generalization as Premise 1 and complete Premise 2 and the Conclusion using the given
subject. Below the arguments, draw a single diagram showing the premises of both arguments
true (two circles, and X1 and X2).

• [Premise 1] No one over age 24 truly understands what it’s like to be a teenager today.

• [Premise 2] My archery instructor (X1)… • [Premise 2] This podcast host (X2)…

_______________________________ _______________________________

_______________________________ _______________________________

_______________________________ _______________________________

[Conclusion] ________________________ [Conclusion] ________________________

__________________________________ __________________________________

__________________________________ __________________________________

Diagram

2) Prove that the given argument is invalid using a labeled diagram with a brief explanation of how
it proves this.

• Every one of the great civilizations of the Diagram

Bronze Age (3300-1200 BCE) Near East
believed in the supremacy of Marduk.
• The Zazakadian Empire was not one of the
great civilizations of the Bronze Age.

The Zazakadian Empire did not believe in the

supremacy of Marduk.

- 38 -
3) For the given universal generalization statement, write the two categories used in the statement
and convert the statement into logically equivalent statements using the given quantifiers. Once
you have assigned letters (A, B) to the categories, you may use either the full labels or just the
letter labels for the rest of your answer.

Only multiplex movie theatres regularly seat large audiences of young people.

Categories

A: ________________________________________________________________________

B: ________________________________________________________________________

Logically Equivalent Statements

All…_______________________________________________________________________

No…_______________________________________________________________________

Give a counterexample to the statement above and say what it proves. Draw a diagram of the
original generalization. Draw a corrected diagram that includes your counterexample (X).

Counterexample (X): _________________________________________________________

X proves that: Some…_________________________________________________________

Diagram of Original Generalization Diagram of Correction (including X)

- 39 -
 PRACTICE EXAM 5

1) Complete two valid syllogisms with different argument patterns. For each argument, use the given
universal generalization as Premise 1 and complete Premise 2 and the Conclusion using the given
subject. Below the arguments, draw a single diagram showing the premises of both arguments
true (two circles, and X1 and X2).

• [Premise 1] Only encrypted messaging apps give the people of Pseudorica access to
international journalism sources.

• [Premise 2] Whisper (X1)…__________ • [Premise 2] Tapper (X2)…__________

_______________________________ _______________________________

_______________________________ _______________________________

[Conclusion] ________________________ [Conclusion] ________________________

__________________________________ __________________________________

__________________________________ __________________________________

Diagram

2) Draw a labeled diagram that proves the given argument is invalid and give a brief explanation.

• No western country has managed to Diagram

get its smoking prevalence below 10%.
• Azmakia is not a western country.

Azmakia has managed to get its smoking

prevalence below 10%.

- 40 -
3) For the given universal generalization statement, write the two categories used in the statement
and convert the statement into logically equivalent statements using the given quantifiers.
Once you have assigned letters (A, B) to the categories, you may use either the full labels or just
the letter labels for the rest of your answer.

All mental illnesses begin during the teenage years.

Categories

A: ________________________________________________________________________

B: ________________________________________________________________________

Logically Equivalent Statements

Only…_____________________________________________________________________

No…_______________________________________________________________________

Give a counterexample to the statement above and say what it proves. Draw a diagram of the
original generalization. Draw a corrected diagram that includes your counterexample (X).

Counterexample (X): _________________________________________________________

X proves that: Some…________________________________________________________

Diagram of Original Generalization Diagram of Correction (including X)

- 41 -
 -ANSWER KEY-

QUICK TEST
1) d 2) d 3) b 4) a 5) c 6) d 7) a 8) a

PRACTICE EXAM 1

1) Answer
• [Premise 1] Every legitimate election offers voters mail-in ballots.
• [Premise 2] Azmakia’s election (X1) is • [Premise 2] Pseudorica’s election (X2)
a legitimate election. does not offer voters mail-in ballots.

[Conclusion] Azmakia’s election offers [Conclusion] Pseudorica’s election is not a

voters mail-in ballots. legitimate election.

Elections that
Legitimate Offer Voters
Elections Mail-in Ballots

X2 X1

Explanation (not required for exam answer)

Both arguments are valid but each has a different pattern:

• All A are B. • All A are B. A: Legitimate Elections

• X1 is an A. • X2 is not a B. B: Elections that Offer…
X1: Azmakia’s Election
X1 is a B. X2 is not an A. X2: Pseudorica’s Election

In the diagram, the circles show Premise 1 true and X1 and X2 show Premise 2 of each argument
true. Since each argument is valid, showing its premises true automatically shows its conclusion
true.

- 42 -
2) Answer

X: Constant Coffee

Coffee Shops
Businesses that Follow
that Stay Public Health
Open Rules

X The diagram imagines the conclusion false when

the premises are true. That possibility proves this
deductive argument is invalid.

Explanation (not required for exam answer)

The invalid argument • Only A are B. A: Coffee Shops that Follow…
pattern is: • X is an A. B: Businesses that Stay Open
X: Constant Coffee
X is a B.

3) Answer
A: Industries
B: Things that Extract Natural Resources From the Ground

All industries do not extract natural resources from the ground.

[Also correct: All things that extract natural resources from the ground are not industries.]
[All A are Non-B. All B are Non-A.]
Only things that don’t extract natural resources from the ground are industries.
[Also correct: Only non-industries extract natural resources from the ground.]
[Only Non-B are A. Only Non-A are B.]

Counterexample (X): mining

X proves that: Some industries extract natural resources from the ground.
[Some A are B.]

Original Generalization Correction (including X)

Industries Things that Industries Things that

[A] Extract Natural [A] Extract Natural
Resources from Resources from
X
the Ground [B] the Ground [B]

- 43 -
Explanation (not required for exam answer)
The predicate begins on the verb extract. The category could be more specific than Things that
Extract…. It could be Activities that Extract…, Projects that Extract…, etc. Mining, or some
particular type of mining, is an obvious counterexample but other answers could be correct. This
question is not a test of factual knowledge. For example, Bob digs a well and then sells glasses of
water. Is the water really a natural resource? Is the business really an industry? It doesn’t matter
for this question. The answer shows that you’ve done the logic correctly.

The corrected diagram shows the minimum correction – each category partially but not completely
overlapping the other – that allows placement of X. A diagram showing complete overlap
(containment) is more than necessary but still correct.

- 44 -
 PRACTICE EXAM 2

1) Answer
• [Premise 1] Only Biosafety Level 4 medical labs can safely study samples of Ebola virus.
• [Premise 2] The Gotham Bioresearch • [Premise 2] Azmakia’s National
Institute’s lab (X1) can safely study Virology Lab (X2) is not a Biosafety
samples of Ebola virus. Level 4 lab.

[Conclusion] The Gotham Bioresearch [Conclusion] Azmakia’s National Virology

Institute’s lab is a Biosafety Level 4 lab. Lab cannot safely study samples of Ebola
virus.

Labs that Can

Biosafety
Safely Study Level 4 Labs
Samples of
Ebola Virus
X2 X1

Explanation (not required for exam answer)

Both arguments are valid but each has a different pattern:

• Only A are B. • Only A are B. A: Biosafety Level 4 Labs

• X1 is a B. • X2 is not an A. B: Labs that Can Safely…
X1: Gotham BI lab
X1 is an A. X2 is not a B. X2: Azmakia NV lab

In the diagram, the circles show Premise 1 true and X1 and X2 show Premise 2 of each argument
true. Since each argument is valid, showing its premises true automatically shows its conclusion
true.

2) Answer

X: Iceberg B24c

Icebergs X Icebergs
that Break that Float
Off from into Tropical
Antarctic Ice Waters
Shelf The diagram shows how it’s possible for this
argument’s conclusion to be false even if its
premises are true. An argument like that isn’t
valid.

- 45 -
 Explanation (not required for exam answer)
The invalid argument • No A are B. A: Icebergs that Break Off…
pattern is: • X is not a B. B: Icebergs that Float into…
X: Iceberg B24c
X is an A.

3) Answer
A: Museums
B: Places that Display Artworks

Counterexample (X1): natural history museum

X1 proves that: Some museums do not display artworks. [Some A are not B.]
Not all museums display artworks.

Counterexample (X2): a restaurant with a painting on the wall

X2 proves that: Some places that display artworks are not museums. [Some B are not A.]
Not only museums display artworks.

Original Generalization Correction (including X1 and X2)

Museums [A]
Museums [A] X2
Places that
Places that Display X1 Display
Artworks [B] Artworks [B]

Explanation (not required for exam answer)

The predicate is “display artworks”. The predicate category could be more specific, e.g. Buildings
that Display Artworks. The category you give could affect which things are counterexamples. For
example, a sidewalk (outdoor) art fair is a counterexample if the category is Places that Display
Artworks but not if the category is Buildings that Display Artworks.

The corrected diagram shows the minimum correction – each category partially but not completely
separated from the other – that allows placement of X1 and X2. A diagram showing complete
separation is more than necessary but still correct.

- 46 -
 PRACTICE EXAM 3

1) Answer

• [Premise 1] No companies that are good to invest in have a high-school drop-out CEO
• [Premise 2] Wackadoodle Industries • [Premise 2] Flippers Spatula Co. has a
(X1) is good to invest in. high-school drop-out CEO.

[Conclusion] Wackadoodle Industries does [Conclusion] Flippers Spatula Co. is not a

not have a high-school drop-out CEO. good company to invest in.

Companies with a Companies that

High-School are Good to
Drop-Out CEO Invest In

X2 X1

Explanation (not required for exam answer)

Both arguments are valid but each has a different pattern:

• No A are B. • No A are B. A: Companies that are Good

• X1 is an A. • X2 is a B. to Invest In
B: Companies with a High-
X1 is not a B. X2 is not an A. School Drop-Out CEO
X1: Wackadoodle Industries
X2: Flippers Spatula Co.

In the diagram, the circles show Premise 1 true and X1 and X2 show Premise 2 of each argument
true. Since each argument is valid, showing its premises true automatically shows its conclusion
true.

2) Answer

X: UN’s Trillion Trees Initiative

X
Things that Tree-
Can Prevent Planting The argument is invalid because the X shows a
Climate Programs
Change
way to make the conclusion false and the premises
true. That’s impossible with a valid argument.

- 47 -
 Explanation (not required for exam answer)
The invalid argument • Only A are B. A: Tree-Planting Programs
pattern is: • X is an A. B: Things that Can Prevent…
X: UN’s Trillion Trees Initiative
X is a B.

3) Answer
A: Wildfires
B: Fires that Starts from a Lightning Strike

Only fires that start from a lightning strike are wildfires.

[Also correct: Only non-wildfires do not start from a lightning strike.]
[Only B are A. Only Non-A are Non-B.]
No wildfires do not start from a lightning strike.
[Also correct: No fires that do not start from a lightning strike are wildfires.]
[No A are Non-B. No Non-B are A.]

Counterexample (X): wildfire that starts from a tossed cigarette

X proves that: Some wildfires do not start from a lightning strike. [Some A are not B.]

Original Generalization Correction (including X)

Fires that Start from a

Lightning Strike [B] Wildfires [A]
Fires that Start
from a
Lightning Strike
Wildfires [A] X [B]

Explanation (not required for exam answer)

The predicate begins on the verb starts.

The corrected diagram shows the minimum correction – the categories partially but not completely
separated – that allows placement of X. A diagram showing complete separation is also correct.

- 48 -
 PRACTICE EXAM 4

1) Answer
• [Premise 1] No one over age 24 truly understands what it’s like to be a teenager today.
• [Premise 2] My archery instructor (X1) • [Premise 2] This podcast host (X2)
is over age 24. truly understands what it’s like to be a
teenager today.
[Conclusion] My archery instructor does
not truly understand what it’s like to be a [Conclusion] This podcast host is not over
teenager today. age 24.

People Over People Who Truly

Understand What It’s
Age 24
Like to be a Teenager
Today
X1 X2

Explanation (not required for exam answer)

Both arguments are valid but each has a different pattern:

• No A are B. • No A are B. A: People Over Age 24

• X1 is an A. • X2 is a B. B: People Who Truly…
X1: My archery instructor
X1 is not a B. X2 is not an A. X2: This podcast host

In the diagram, the circles show Premise 1 true and X1 and X2 show Premise 2 of each argument
true. Since each argument is valid, showing its premises true automatically shows its conclusion
true.

2) Answer

X: Zazakadian Empire

X
Civilizations that
Believed in the
Great Bronze
Supremacy of
Age
Civilizations
Marduk The location of the X makes the
conclusion false and the premises true.
Any deductive argument for which you
can do that is invalid.

- 49 -
 Explanation (not required for exam answer)
The invalid argument • All A are B. A: Great Bronze Age…
pattern is: • X is not an A. B: Civilizations that Believed…
X: Zazakadian Empire
X is not a B.

3) Answer
A: Multiplex Movie Theatres
B: Places that Regularly Seat Large Audiences of Young People

All places that regularly seat large audiences of young people are multiplex movie theatres.
[All B are A.]
No places that regularly seat large audiences of young people are not multiplex movie theatres.
[No B are Non-A.]

Counterexample (X): school lecture hall

X proves that: Some places that regularly seat large audiences of young people are not multiplex
movie theatres. [Some B are not A.]

Original Generalization Correction (including X)

Places that Multiplex

Multiplex Movie Theatres [A]
Regularly Seat Movie
Large Audiences Theatres
of Young People [A]
Places that
[B]
Regularly Seat
Large Audiences of X
Young People [B]

Explanation (not required for exam answer)

The predicate begins on the adverb regularly. The category could be more specific than Places
that Regularly…. It could be Building that Regularly…, etc. A school lecture hall is an obvious
counterexample but anything that could reasonably be described the same way would be correct.

The corrected diagram shows the minimum correction – the categories partially but not completely
separated – that allows placement of X. A diagram showing complete separation is also correct.

- 50 -
 PRACTICE EXAM 5

1) Answer
• [Premise 1] Only encrypted messaging apps give the people of Pseudorica access to
international journalism sources.
• [Premise 2] Whisper (X1) gives the • [Premise 2] Tapper (X2) is not an
people of Pseudorica access to encrypted messaging app.
international journalism sources.
[Conclusion] Tapper does not give the
[Conclusion] Whisper is an encrypted people of Pseudorica access to
messaging app. international journalism sources.

Things that Give Encrypted

the People of
Pseudorica Access Messaging Apps
to International
Journalism
Sources
X2 X1

Explanation (not required for exam answer)

Both arguments are valid but each has a different pattern:

• Only A are B. • Only A are B. A: Encrypted Messaging Apps

• X1 is a B. • X2 is not an A. B: Things that Give…
X1: Whisper
X1 is an A. X2 is not a B. X2: Tapper

In the diagram, the circles show Premise 1 true and X1 and X2 show Premise 2 of each argument
true. Since each argument is valid, showing its premises true automatically shows its conclusion
true.

2) Answer

X: Azmakia
Western
X Countries that
Countries Manage to Get
the Smoking
Prevalence We can imagine this deductive argument’s
Below 10%
conclusion false and its premises true at the same
time. That’s possible only if it’s invalid.

- 51 -
 Explanation (not required for exam answer)
The invalid argument • No A are B. A: Western Countries
pattern is: • X is not an A. B: Countries that Manage…
X: Azmakia
X is a B.

3) Answer
A: Mental Illnesses
B: Things that Begin During the Teenage Years

Only things that begin during the teenage years are mental illnesses.
[Also correct: Only things that aren’t mental illnesses do not begin during the teenage years.
[Only B are A. Only Non-A are Non-B.]
No mental illnesses do not begin during the teenage years.
[Also correct: No things that don’t begin during the teenage years are mental illnesses.]
[No A are Non-B. No Non-B are A.]

Counterexample (X): post-traumatic stress disorder from adult war experiences

X proves that: Some mental illnesses do not begin during the teenage years. [Some A are not B.]

Original Generalization Correction (including X)

Things that Things that

Begin During Mental
Mental Begin During
Illnesses
a Person’s a Person’s
Illnesses [A]
Teenage Teenage
[A]
Years [B] Years [B]
X

Explanation (not required for exam answer)

The predicate begins on the verb begin. The predicate category could be more specific than Things
that Begin…, e.g. Illnesses that Begin…. Anything that could reasonably be described the same
way as this counterexample would be correct.

The corrected diagram shows the minimum correction – the categories partially but not completely
separated from each other – that allows placement of X. A diagram showing complete separation
is also correct.

- 52 -
 ~3~
-STATEMENT LOGIC-

i. -LOGICAL OPERATORS-

Simple statements (represented by letters: P, Q, etc.) may be modified or connected with a logical
operator, which attaches to the statements and “operates” on their truth values. This forms a new
statement whose truth value is determined by the truth value(s) of the simple statement(s).

For example, if P is true, its negation ¬P is false; if P is false, Negation

its negation ¬P is true. P is logically equivalent to the negation ¬P
of its negation: P 𠪪P. Says: NOT P (P is false.)

The conjunction and disjunction operators create conjunction and disjunction statements,
compound statements that connect two simple statements.

Conjunction Disjunction
P-AND-Q Inclusive Exclusive
Says: BOTH P and Q are true. P-OR-Q P-XOR-Q
Says: AT LEAST ONE Says: EXACTLY ONE
of P and Q is true. of P and Q is true.

A four-line truth table shows the four possible truth value combinations for two simple statements.
Along each line, it shows the resulting truth values of negation, conjunction, and the disjunctions.

P Q ¬P P-AND-Q P-OR-Q P-XOR-Q

1 True True False True True False
2 True False False False True True
3 False True True False True True
4 False False True False False False


We’ll use the convenient symbol ¬ instead of NOT, and we’ll use AND and OR instead of the symbols ∧ and ∨.

- 53 -
ii. -DISJUNCTIVE SYLLOGISM-

In a disjunctive syllogism, one premise says that one of two disjuncts is true, and the other
premise denies one disjunct. The conclusion is the other disjunct.

Disjunctive Syllogism (DS)

• P-OR-Q
• ¬P

Argument 3.1
• Insulin (a hormone that allows the body to use
sugar) is released into the bloodstream from the
kidneys or from the liver.
• It isn’t released from the liver.

It is released from the kidneys.

(Image: NIDDK)

Simple Statements Argument 3.1 Pattern

P: Insulin is released from the • P-OR-Q
kidneys. • ¬Q
Q: Insulin is released from the liver.
P
VALID (DS)

The argument has a false conclusion (insulin actually comes from the pancreas) even though it’s valid.
That’s possible because the argument has a false premise. (The premise P-OR-Q is false because both
P and Q are false.)

Does the disjunction premise in Argument 3.1 seem exclusive (XOR) or inclusive (OR) to you? Ordinary
language often doesn’t make this clear. (People sometimes write “and/or” to make clear they mean
OR.) We’ll normally read or as OR, but for DS arguments such as Argument 3.1, it doesn’t matter.
Both OR and XOR say “one of these two statements is true”. So if one statement is false, the other
statement is true. So DS is valid with either type of disjunction.

- 54 -
iii. -CONDITIONAL-

A conditional is an If…then… compound statement. Unlike conjunction and disjunction, the truth
value of a conditional is not determined by the truth values of the statements within it. A conditional
is true when the truth of its antecedent statement is sufficient for (guarantees) the truth of its
consequent, and the truth of its consequent is necessary (required) for the truth of its antecedent.

Conditional
If [Antecedent] then [Consequent].
Says: [Antecedent] SUFFICIENT for [Consequent].
[Consequent] NECESSARY for [Antecedent].

Conditional 3.2: If whales are fish then they can

breathe underwater.
Antecedent: Whales are fish.
Consequent: Whales can breathe underwater.

Often people put a question or command into an If…then… sentence instead of a consequent
statement. A sentence like this can easily be re-written as a proper conditional.

Conditional 3.3: If there are other technologically

advanced beings in the galaxy then where are they?
If there were other technologically advanced beings in
the galaxy then we would have encountered them.

Conditional 3.4: If you want to call yourself a male

feminist, demand that your female colleagues be paid the
same as you are paid.
If you want to call yourself a male feminist, you should
demand that your female colleagues be paid the same as
you are paid.
(Image: bizjournals.com)


In a course on formal symbolic logic, there is a logical operator for If…then…. This type of conditional works a bit differently
from the one we’ve described here.

- 55 -
 Writing Conditionals

Conditionals are often expressed in one of three different logically equivalent forms. The word when
functions logically as if, so any of these forms may also be written with when instead of if.

Conditional Forms
If [Antecedent] then [Consequent].
[Consequent] if [Antecedent].
[Antecedent] only if [Consequent].

Conditional 3.5: Tennis was invented before golf if the

ancient Egyptians played tennis.
Antecedent: The ancient Egyptians played tennis.
Consequent: Tennis was invented before golf.

Conditional 3.6: Bob will get his passport in time for his trip only
if he pays the extra fee for fast processing.
Antecedent: Bob gets his passport in time for his trip.
Consequent: Bob pays the extra fee for fast processing.

In Conditional 3.6, only if emphasizes that paying the extra fee is necessary for Bob to get his passport.

Unless

An unless sentence is also a conditional. “P unless Q” says that P is true if Q is false: P if ¬Q.

Statement 3.7: You’ll hurt yourself exercising unless you

stretch first.
You’ll hurt yourself exercising if you do not stretch first.
Antecedent: You do not stretch before exercising.
Consequent: You’ll hurt yourself exercising.

- 56 -
-CONDITIONAL ARGUMENTS-

Hypothetical Syllogism

A hypothetical syllogism has two conditional premises. These overlap on a statement that is
antecedent in one conditional and consequent in the other. The conclusion is the conditional we get
by cutting out the overlapping statement.

Hypothetical Syllogism (HS)

• If P then Q.
• If Q then R.

If P then R.

Argument 3.8
• There will be a huge oil spill if that oil tanker crashes.
• If there’s a huge oil spill then many of the animals that live in this area will not survive.

Many of the animals that live in this area will not survive if that oil tanker crashes.

Simple Statements Argument 3.8 Pattern

P: That oil tanker crashes. • R if P.
Q: The animals will survive. • If R then ¬Q.
R: There is a huge oil spill.
¬Q if P.
VALID (HS)

- 57 -
 Affirming Antecedent (AA) and the AC Fallacy

“If P then Q” says P is sufficient for Q. It does not say Q is sufficient for P. This means that:

Affirming Antecedent (AA) Affirming Consequent (AC)

• If P then Q. • If P then Q.
• P • Q

Q P

VALID INVALID (Fallacy)

Argument 3.9 (AA)

• If food prices rise, people will riot.
• Food prices are rising.

(Image: Food & Beverage Insider)

People will riot.

Argument 3.10 (AC Fallacy)

• People riot when food prices have risen.
• People are rioting.

Food prices have risen.

(Image: Nosyrevy)

Simple Statements Argument 3.9 Pattern Argument 3.10 Pattern

P: Food prices rise. • If P then Q. • Q if P.
Q: People riot. • P • Q

Q P
VALID (AA) INVALID (AC)

Argument 3.10 is a logical fallacy; it is invalid because its conclusion could be false even when its
premises are true. For example, perhaps there’s a riot because a local sports team lost a big game
instead of food prices.

- 58 -
 Denying Consequent (DC) and the DA Fallacy

“If P then Q” says Q is necessary for P. It does not say P is necessary for Q. This means that:

Denying Antecedent (DA) Denying Consequent (DC)

• If P then Q. • If P then Q.
• ¬P • ¬Q

¬Q ¬P

INVALID (Fallacy) VALID

Argument 3.11 (DC)

• MDMA (Ecstasy) would be legal unless the public
believed that using it is a sign of bad personal character.
• MDMA is not legal.

The public does believe that.

Argument 3.12 (DA Fallacy)

• MDMA would be legal only if the public didn’t believe
that using it is a sign of bad personal character.
• MDMA is not legal.

(Image: South Park) The public does believe that.

Simple Statements Argument 3.11 Pattern Argument 3.12 Pattern

P: Recreational MDMA is legal. • P if ¬Q. • P only if ¬Q.
Q: Public believes MDMA is sign of • ¬P • ¬P
bad character.
Q Q
VALID (DC) INVALID (DA)

Argument 3.14 is invalid because its conclusion could be false even when its premises are true.
Perhaps the public doesn’t believe that using MDMA is a sign of bad character even though it’s illegal.

- 59 -
 DC for Objections

Bob says that eating mammals is morally OK. Abby disagrees. She objects to (against) his statement
with a DC argument. To make this argument, she needs a conditional with three features:
1) Bob’s statement is the antecedent.
2) The whole conditional is true, so the conditional premise of the argument is true.
3) The consequent is false, so the DC premise of the argument is true.

Objection against

Eating mammals is morally OK only if they do not suffer in factory farms.

False Consequent

Following the DC pattern, Abby concludes that the antecedent (Bob’s statement) is false.

Argument 3.13
• Eating mammals is morally OK only if they do not
suffer in factory farms.
• They do suffer in factory farms.

Eating mammals is not morally OK.

(Image: Animal Outlook)

Simple Statements Argument 3.13 Pattern

P: Eating mammals is OK. • P only if ¬Q.
Q: Mammals suffer in factory farms. • Q

¬P
VALID (DC)

Abby’s objection is a valid argument. If Bob continues to believe that eating mammals is morally OK,
he must reject one of her premises. To reject the conditional premise, he must say that the moral OK-
ness of eating mammals really does not require that they not suffer in factory farms. To reject the DC
premise, he must say that mammals do not suffer in factory farms.

- 60 -
 AA and DC with Complex Conditionals

A complex conditional has a compound statement (AND or OR) in it. A complex conditional premise
can make it a bit more difficult to see whether a conditional argument is a successful AA or DC pattern.

Argument 3.14
• If this news report is true and Ladnod Prumt is an
honourable politician, he will resign.
• This news report is definitely true.

Prumt will resign.

Simple Statements Argument 3.14 Pattern

P: This news report is true. • If P-AND-Q then R.
Q: Prumt is an honourable politician. • P
R: Prumt will resign.
R
INVALID

The argument resembles AA, but it’s invalid. The antecedent P-AND-Q is true when both P and Q
are true. But the other premise affirms only P. Perhaps Q is false.

AC and DA in Everyday Language

Once you’re aware of the AC and DA patterns, you will notice them everywhere.

Argument 3.15: I think our restaurant won’t be able to stay

in business. It would be able to if customers took selfies for
their social media, but they don’t do that.

Argument 3.15 looks like a DA fallacy. But perhaps it has an implicit biconditional premise: “Our
restaurant would be able to stay in business if and only if customers took selfies for their social media.”
Then it would be valid (DC). Or maybe the premises are not supposed to guarantee the conclusion
but only to make it likely. We’ll see some non-deductive argument patterns like this in Unit 4.

Similarly, many apparent examples of AC may actually be either implicitly valid or non-deductive.

- 61 -
 -UNIT 3 SKILLS-

You must be able to:

• Recognize and write negations, conjunctions, and disjunctions.

• Recognize and write conditionals in three forms, plus unless.

• Construct the valid syllogisms DS, HS, AA, and DC.

• Determine whether a conditional argument with a complex conditional is valid or invalid (successful

or failed AA or DC).

• Construct an objection against a statement using the DC pattern.

- 62 -
 -QUICK TEST QUESTIONS-

1) What does a conjunction of two statements say?

a) Neither statement is true. c) One statement is true and the other is false.
b) Both statements are true. d) At least one statement is true.

2) What does a disjunction of two statements say?

a) Neither statement is true. c) One statement is true.
b) One statement could be true. d) Both statements are true.

3) “If P then Q” says:

a) Q is the antecedent and P is the c) P is the antecedent and Q is the
consequent; P is sufficient for Q. consequent; Q is sufficient for P.
b) P is the antecedent and Q is the d) Q is the antecedent and P is the
consequent; P is sufficient for Q. consequent; Q is sufficient for P.

4) Which conditional form emphasizes that Q is necessary for P?

a) If P then Q. c) Q when P.
b) Q if P. d) P only if Q.

5) “P if Q, but ¬Q. Therefore ¬P.” This pattern is:

a) DA c) DC
b) AA d) AC

6) Which are valid syllogism patterns?

a) DC and AC c) AA and DA
b) AC and DA d) AA and DC

7) When DC is used to make an objection against a statement P, where is P in the conditional

premise?
a) Antecedent: P c) Consequent: P
b) Antecedent: ¬P d) Consequent: ¬P

- 63 -
 -PRACTICE EXAMS-

PRACTICE EXAM 1

1) Complete two different disjunctive syllogisms (DS) using the given Premise 1. You may change
tense (past/present/future) in your answers.

• [Premise 1] Either Findley Yardman (the famous actor) will go to jail for drunk driving or he’ll
lose his remaining money at the casino.

• [Premise 2] _____________________ • [Premise 2] _____________________

_________________________________ _______________________________

[Conclusion] _______________________ [Conclusion] ________________________

__________________________________ __________________________________

2) Complete a reasonable hypothetical syllogism (HS) using the given Premise 1.

• [Premise 1] Unless Azmakia protects its coastal wetlands, the blue-tailed warbler will not have
a place to nest.
• [Premise 2] _______________________________________________________________

[Conclusion] __________________________________________________________________

3) Write a DC argument as an objection against the statement. For full marks, make the objection
reasonable. You may change the tense (past/present/future) of the statement.

Teenagers should be allowed to have e-bikes.

4) Is this argument: valid (AA), valid (DC), invalid (failed AA), or invalid (failed DC)?

• If people cared about their privacy, they would lock their doors and they would not share
every detail of their life on social media.
• People do share every detail of their life on social media.

People do not care about their privacy.

- 64 -
 PRACTICE EXAM 2

1) Complete AA and DC syllogisms using the given Premise 1. You may change tense
(past/present/future) in your answers.

• [Premise 1] The president resigns only if the people do not still support her.

• [Premise 2] _____________________ • [Premise 2] _____________________

_______________________________ _______________________________

[Conclusion] _______________________ [Conclusion] ________________________

_________________________________ __________________________________

2) Complete a hypothetical syllogism (HS) using the given Premise 1. For full marks, write reasonable
premises.

• [Premise 1] If Netflix stops people from sharing passwords then we’ll have to watch YouTube.

• [Premise 2] ________________________________________________________________

[Conclusion] __________________________________________________________________

3) Write a DC argument as an objection against the statement. For full marks, make the objection
reasonable. You may change the tense (past/present/future) of the statement.

The revolution succeeds.

4) Is this argument: valid (AA), valid (DC), invalid (failed AA), or invalid (failed DC)?

• We’ll see a Swaylor Tift concert next year if we’re still fans of her then and we can afford the
tickets.
• We’ll still be fans of her next year.

We’ll see her perform live next year.

- 65 -
 PRACTICE EXAM 3

5) Complete two different disjunctive syllogisms (DS) using the given Premise 1. You may change
tense (past/present/future) in your answers.

• [Premise 1] Either nuclear fusion is a viable energy source or civilization as we know it does
not have a long future on Earth.

• [Premise 2] _____________________ • [Premise 2] _____________________

_________________________________ _______________________________

[Conclusion] _______________________ [Conclusion] ________________________

__________________________________ __________________________________

6) Complete a hypothetical syllogism (HS) using the given Premise 1. For full marks, write reasonable
premises.

• [Premise 1] Bob will get a different apartment if he’s afraid of heights.

• [Premise 2] _______________________________________________________________

[Conclusion] __________________________________________________________________

7) Write a DC argument as an objection against the statement. For full marks, make the objection
reasonable. You may change the tense (past/present/future) of the statement.

Video games should not be an Olympic sport.

8) Is this argument: valid (AA), valid (DC), invalid (failed AA), or invalid (failed DC)?

• The central bank would raise the interest rate only if inflation is too high and there’s risk of a
housing bubble.
• The central bank raised the interest rate.

There’s risk of a housing bubble.

- 66 -
 PRACTICE EXAM 4

1) Complete AA and DC syllogisms using the given Premise 1. You may change tense
(past/present/future) in your answers.

• [Premise 1] Gun sales go up whenever people don’t trust the government to maintain social
order.

• [Premise 2] ____________________ • [Premise 2] _____________________

_______________________________ _______________________________

[Conclusion] _______________________ [Conclusion] ________________________

__________________________________ __________________________________

2) Complete a hypothetical syllogism (HS) using the given Premise 1. For full marks, write a
reasonable Premise 2.

• [Premise 1] Free speech is worth protecting unless people say things that make me angry.

• [Premise 2] _________________________________________________________________

[Conclusion] ___________________________________________________________________

3) Write a DC argument as an objection against the statement. For full marks, make the objection
reasonable.

Hamsters are the best pets.

4) Is this argument: valid (AA), valid (DC), invalid (failed AA), or invalid (failed DC)?

• If the rebel groups had formed an alliance, attacks on government forces would have become
bolder or more frequent.
• Attacks on government forces have not become more frequent.

The rebel groups did not form an alliance.

- 67 -
 PRACTICE EXAM 5

1) Complete AA and DC syllogisms using the given Premise 1. You may change tense
(past/present/future) in your answers.

• [Premise 1] There will be a World Cup this year unless the pandemic doesn’t end.

• [Premise 2] ____________________ • [Premise 2] _____________________

_______________________________ _______________________________

[Conclusion] _______________________ [Conclusion] ________________________

__________________________________ __________________________________

2) Complete a hypothetical syllogism (HS) using the given Premise 1. For full marks, write a
reasonable Premise 2 and Conclusion.

• [Premise 1] Humans will travel to Mars only if national space programs get more funding.

• [Premise 2] _________________________________________________________________

[Conclusion] ___________________________________________________________________

3) Write a DC argument as an objection against the statement. For full marks, make the objection
reasonable.

People will never get bored of superhero movies.

4) Is this argument: valid (AA), valid (DC), invalid (failed AA), or invalid (failed DC)?

• There are summer wildfires in this area whenever there’s no rain in the spring or there’s been
a recent heatwave.
• There are no summer wildfires in this area.

There was rain in the spring.

- 68 -
 -ANSWER KEY-

QUICK TEST
1) b 2) c 3) b 4) d 5) a 6) d 7) a

PRACTICE EXAM 1

1) Answer

• [Premise 1] Either Findley Yardman (the famous actor) will go to jail for drunk driving or he’ll
lose his remaining money at the casino.
• [Premise 2] He doesn’t go to jail for • [Premise 2] He doesn’t lose his
drunk driving. remaining money at the casino.

[Conclusion] He loses his remaining [Conclusion] He goes to jail for drunk

money at the casino. driving.

Explanation (not required for exam answer)

The argument patterns are:

• P or Q. • P or Q. P: FY goes to jail for drunk

• ¬P • ¬Q driving.
Q: FY loses his remaining
Q P money at the casino.

2) Answers (examples)

• [Premise 1] Unless Azmakia protects its coastal wetlands, the blue-tailed warbler will not have
a place to nest.
• [Premise 2] Unless the citizens of Azmakia appreciate the wildlife in the coastal wetlands,
Azmakia will not protect its coastal wetlands.

[Conclusion] Unless the citizens of Azmakia appreciate the wildlife in the coastal wetlands, the
blue-tailed warbler will not have a place to nest.

- 69 -
 • [Premise 1] Unless Azmakia protects its coastal wetlands, the blue-tailed warbler will not
have a place to nest.
• [Premise 2] Unless the blue-tailed warbler has a place to nest, it will go extinct.

[Conclusion] Unless Azmakia protects its coastal wetlands, the blue-tailed warbler will go extinct.

Explanation (not required for exam answer)

Since the given conditional Premise 1 contains two simple statements, an answer must contain a
third simple statement to make an HS argument. In the first answer, Premise 2 connects with the
antecedent of Premise 1 (“Azmakia will not protect its coastal wetlands”); in the second answer,
Premise 2 connects with the consequent of Premise 1 (“The blue-tailed warbler will not have a
place to nest”).

3) Answers (examples)

• Teenagers should be allowed to have e-bikes only if they will not modify them in order to
exceed safe riding speeds.
• Teenagers will modify e-bikes in order to exceed safe riding speeds.

Teenagers should not be allowed to have e-bikes.

• If teenagers should be allowed to have e-bikes then they should be allowed to have
motorcycles.
• Teenagers should not be allowed to have motorcycles.

Teenagers should not be allowed to have e-bikes.

Explanation (not required for exam answer)

The argument is an objection to the statement “Teenagers should be allowed to have e-bikes”.
For the DC argument pattern, this statement is the antecedent of the conditional premise. The
conclusion denies this statement: Teenagers should not be allowed to have e-bikes.

4) Answer

Valid (DC)

- 70 -
Explanation (not required for exam answer)
The argument pattern is:

• If P then Q-AND-(¬R). P: People care about their privacy.

• R Q: People lock their doors.
R: People share every detail of their life on
¬P social media.

The consequent is a conjunction Q-AND-(¬R), which is true when Q is true and R is false. So this
consequent may be denied by saying that Q is false, or that R is true, or both of these. The
premise does this, so the conclusion ¬P is guaranteed by the premises.

- 71 -
 PRACTICE EXAM 2

5) Answer

• [Premise 1] The president resigns only if the people do not still support her.
• [Premise 2] The president resigns. • [Premise 2] The people do support the
president still.
[Conclusion] The people do not still
support her. [Conclusion] She doesn’t resign.
[AA] [DC]

Explanation (not required for exam answer)

The argument patterns are:

• P only if ¬Q. • P only if ¬Q. P: President resigns.

• P • Q Q: People support the
president.
¬Q ¬P

6) Answers (examples)

• [Premise 1] If Netflix stops people from sharing passwords then we’ll have to watch YouTube.
• [Premise 2] If Netflix profits go down then they will stop people from sharing passwords.

[Conclusion] If Netflix profits go down then we’ll have to watch YouTube.

• [Premise 1] If Netflix stops people from sharing passwords then we’ll have to watch YouTube.
• [Premise 2] If we have to watch YouTube then we’ll be forced to watch lots of ads.

[Conclusion] If Netflix stops people from sharing passwords then we’ll be forced to watch lots of
ads.

Explanation (not required for exam answer)

Since the given conditional Premise 1 contains two simple statements, an answer must contain a
third simple statement to make an HS argument. In the first answer, Premise 2 connects with the
antecedent of Premise 1 (“Netflix stops people from sharing passwords”); in the second answer,
Premise 2 connects with the consequent of Premise 1 (“We’ll have to watch YouTube”).

- 72 -
7) Answers (examples)

• The revolution will succeed only if the government doesn’t shut down the internet that
revolutionaries use to coordinate their protests.
• The government is shutting down the internet that revolutionaries use to coordinate their
protests.

The revolution will not succeed.

• The political prisoners would have been released by now if the revolution had succeeded.
• The political prisoners still have not been released.

The revolution failed. [see explanation below]

• If the revolution were succeeding, the army would be joining with the protestors against the
government.
• The army is not joining with the protestors against the government.

The revolution is not succeeding.

Explanation (not required for exam answer)

The argument is an objection to the statement “The revolution succeeds”, in the future, past, or
present tense. For the DC argument pattern, this statement is the antecedent of the conditional
premise. The conclusion denies this statement: The revolution does not succeed. This situation is
imaginary, but it’s reasonable to think the consequent might be necessary for the antecedent.

(Note: Sometimes there is a word that clearly means the negation of another word. In the second
answer here, fail simply means “not succeed”. In cases like this where the meaning is completely
clear, a correct answer can use a word substitution instead of adding or removing not.)

8) Answer

Invalid (failed AA)

- 73 -
Explanation (not required for exam answer)
The argument pattern is:

• P if Q-AND-R. P: We’ll see a Swaylor Tift concert next year.

• Q Q: We’re still fans of her then.
R: We can afford the tickets.
P

The antecedent is a conjunction Q-AND-R, which is true only when both Q and R are true. To
affirm that antecedent, a premise must affirm both Q and R. In this argument, the premise affirms
only Q. R might be false. So the argument is not a successful AA pattern. It is invalid.

- 74 -
 PRACTICE EXAM 3

1) Answer

• [Premise 1] Either nuclear fusion is a viable energy source or civilization as we know it does
not have a long future on Earth.
• [Premise 2] Nuclear fusion is not a • [Premise 2] Civilization as we know it
viable energy source. has a long future on Earth.

[Conclusion] Civilization as we know it [Conclusion] Nuclear fusion is a viable

does not have a long future on Earth. energy source.

Explanation (not required for exam answer)

Since the disjunct “Civilization as we know it does not have a long future on Earth” already contains
the NOT operator, its negation in the Conclusion (left argument) and Premise 2 (right argument)
removes the word not.
The argument patterns are:

• P OR ¬Q. • P OR ¬Q. P: Nuclear fusion is viable.

• ¬P • Q Q: Civilization has a long
future.
¬Q P

2) Answer (example)

• [Premise 1] Bob isn’t able to go near the window in his apartment if he’s afraid of heights.
• [Premise 2] Bob will get a different apartment if he isn’t able to go near the window in it.

[Conclusion] Bob will get a different apartment if he’s afraid of heights.

Explanation (not required for exam answer)

Since the given conditional Conclusion contains two simple statements, an answer must contain
a third simple statement to make an HS argument. In the HS pattern, both the antecedent and
the consequent of the conclusion are used in the premises. Premise 1 connects with the
antecedent (“Bob’s afraid of heights”); Premise 2 connects with the consequent (“Bob will get a
different apartment”).

- 75 -
3) Answer (example)

• Video games should not be an Olympic sport only if including them in the Olympics will lead to
demands for a bunch of crazy new Olympic sports (whistling, flying paper airplanes, etc.).
• Including video games in the Olympics will not lead to demands for a bunch of crazy new
Olympic sports.

Video games should be an Olympic sport.

Explanation (not required for exam answer)

The argument is an objection to the statement “Video games should not be an Olympic sport”.
For the DC argument pattern, this statement is the antecedent of the conditional premise. The
conclusion denies this statement: Video games should be an Olympic sport. This situation is
imaginary, but it’s reasonable to think the consequent might be necessary for the antecedent.

4) Answer

Valid (AA)

Explanation (not required for exam answer)

The argument pattern is:

• P only if Q-AND-R. P: The central bank raises the interest rate.

• P Q: Inflation is too high.
R: There’s risk of a housing bubble.
R

The premise P affirms the antecedent, which entails the consequent, the conjunction Q-AND-R.
When that conjunction is true, both Q and R are true. So R (the conclusion) is guaranteed by the
premises. (Q is also guaranteed, but the conclusion doesn’t need to say that for the argument to
be valid.)

- 76 -
 PRACTICE EXAM 4

1) Answer

• [Premise 1] Gun sales go up whenever people don’t trust the government to maintain social
order.
• [Premise 2] People don’t trust the • [Premise 2] Gun sales are not going up.
government to maintain social order.
[Conclusion] People trust the government
[Conclusion] Gun sales are going up. to maintain social order.
[AA] [DC]

Explanation (not required for exam answer)

Since the antecedent “People don’t trust the government to maintain social order” already contains
the NOT operator, its negation in the DC Conclusion removes the word not.
The argument patterns are:
• P if ¬Q. • P if ¬Q. P: Gun sales go up.
• ¬Q • ¬P Q: People trust the
government to maintain
P Q social order.

2) Answer (examples)

• [Premise 1] Free speech is worth protecting unless people say things that make me angry.
• [Premise 2] People don’t say things that make me angry unless they make jokes about my
religion.

[Conclusion] Free speech is worth protecting unless people make jokes about my religion.

• [Premise 1] Free speech is worth protecting unless people say things that make me angry.
• [Premise 2] I’ll go to the protest rally downtown unless free speech is not worth protecting.

[Conclusion] I’ll go to the protest rally downtown unless people say things that make me angry.

Explanation (not required for exam answer)

Since the given conditional Premise 1 contains two simple statements, an answer must contain a
third simple statement to make an HS argument. In the HS pattern, both the antecedent and the

- 77 -
 consequent of the conclusion are used in the premises. In the first answer, Premise 2 connects
with the antecedent of Premise 1 (“People don’t say things that make me angry”); in the second
answer, Premise 2 connects with the consequent of Premise 1 (“Free speech is worth protecting”).

3) Answer (examples)

• If hamsters were the best pets, they would fetch a stick and bring it back to you.
• Hamsters won’t fetch a stick and bring it back to you .

Hamsters are not the best pets.

• Hamsters are the best pets only if they don’t bite little children.
• Hamsters do bite little children.

Hamsters are not the best pets.

Explanation (not required for exam answer)

The argument is an objection to the statement “Hamsters are the best pets”. For the DC argument
pattern, this statement is the antecedent of the conditional premise. The conclusion denies this
statement: Hamsters are not the best pets.

4) Answer

Invalid (failed DC)

Explanation (not required for exam answer)

The argument pattern is:

• If P, Q-OR-R. P: The rebel groups formed an alliance.

• ¬R Q: Attacks on government forces become
bolder.
¬P R: Attacks on government forces become more
frequent.

The consequent is Q-OR-R, which is false only when both Q and R are false. The premise ¬R is
not enough to deny this consequent. Perhaps Q is true. So this is not a successful DC pattern. It
is invalid.

- 78 -
 PRACTICE EXAM 5

1) Answer

• [Premise 1] There will be a World Cup this year unless the pandemic doesn’t end.
• [Premise 2] The pandemic ends. • [Premise 2] There won’t be a World
Cup this year.
[Conclusion] There will be a World Cup
this year. [Conclusion] The pandemic isn’t ending.
[AA] [DC]

Explanation (not required for exam answer)

The conditional uses unless ≡ if not, so the antecedent is the double negation “The pandemic does
not not end” ≡ “The pandemic ends”.
The argument patterns are:
• P if ¬¬Q. • P if ¬¬Q. P: There is a World Cup this
• Q • ¬P year.
Q: The pandemic ends.
P ¬Q

2) Answer (examples)

• [Premise 1] Humans will travel to Mars only if national space programs get more funding.
• [Premise 2] National space programs will get more funding only if nations are not preoccupied
with war and disasters.

[Conclusion] Humans will travel to Mars only if nations are not preoccupied with war and disasters.

• [Premise 1] Humans will travel to Mars only if national space programs get more funding.
• [Premise 2] Humans will build permanent settlements beyond Earth only if they travel to Mars.

[Conclusion] Humans will build permanent settlements beyond Earth only if they travel to Mars.

Explanation (not required for exam answer)

Since the given conditional Premise 1 contains two simple statements, an answer must contain a
third simple statement to make an HS argument. In the HS pattern, both the antecedent and the
consequent of the conclusion appear in the premises. In the first example answer, Premise 2

- 79 -
 connects with the consequent of Premise 1 (“National space programs get more funding”); in the
second example answer, Premise 2 connects with the antecedent of Premise 1 (“Humans will
travel to Mars”).

3) Answer (example)

• People will never get bored of superhero movies only if creators can think of more interesting
plots.
• Creators cannot think of more interesting plots.

People will get bored of superhero movies.

Explanation (not required for exam answer)

The argument is an objection to the statement “People will never get tired of superhero movies”.
In this statement, never functions as the NOT operator. For the DC argument pattern, this
statement is the antecedent of the conditional premise. The conclusion denies this statement:
People will get bored of superhero movies.

4) Answer

Valid (DC)

Explanation (not required for exam answer)

The argument pattern is:

• P if (¬Q)-OR-R. P: There are summer wildfires in this area.

• ¬P Q: There was rain in the spring.
R: There’s been a recent heatwave.
Q

The premise ¬P denies the consequent, which entails that the antecedent (¬Q)-OR-R is false.
This disjunction is false only when ¬Q is false (Q is true) and R is false. So Q (the conclusion) is
guaranteed by the premises. (¬R is also guaranteed, but the conclusion doesn’t need to say that
in order for the argument to be valid.)

- 80 -
 ~4~
-EXPLANATION AND CONFIRMATION-

i. -SCIENTIFIC REASONING-

Scientific reasoning is not just for scientists doing research. It is for anyone using observations to gain
knowledge. Often it begins with an attempt to explain some observed fact, to say why or how
something happens. A proposed explanation is one type of hypothesis, a statement that we’re not
yet ready to fully believe. A new observation may confirm (support) or disconfirm the hypothesis.
Evidence is any observation, past or new, that helps us judge the likely truth or falsity of a hypothesis.

A Story of Scientific Reasoning

Abby has pizza for dinner, and some is left

over. She goes out for a while and when
she comes back, a piece of pizza is gone.
“Maybe my cat, Mr. Business, ate that
piece of leftover pizza”, she thinks. Then
she thinks of another explanation: “Maybe
my roommate, Zelda, ate the pizza.”

Mr. Business has taken human food from the table before,
but never anything as big as a whole piece of pizza. Abby
knows that Zelda likes pizza. It makes more sense that
Zelda ate it.

But she wants to test her first idea. “If my cat ate the pizza
then probably he won’t want his cat food tonight.” She
prepares Mr. Business’ usual cat food and waits to see if
(Image: @RealGrumpyCat) he wants it.

- 81 -
ii. -INFERENCE TO THE BEST EXPLANATION-

To explain an event, we think of a hypothesis that would make us expect that event: H explains E
(after we observe E) if we could have used H to predict E (before we observe E). Abby observes that
the pizza is missing. Her hypothesis (“the cat ate it”) explains the missing pizza because it would lead
us to expect the pizza to be gone. Often an explanation is a causal hypothesis, a story about what
caused the event we’re explaining. The cat eating the pizza would cause the pizza to be gone.

Abby thinks of different possible explanations and chooses the best one. This is inference to the
best explanation (IBE). IBE is non-deductive, or inductive, reasoning. Inductive arguments are not
supposed to be valid – instead they are evaluated as strong or weak. In a strong argument, the
premises support the conclusion well but don’t guarantee it.

Why is IBE inductive rather than deductive? There are a few reasons:
• There is no logical method for determining that one explanation is better than another.
• Maybe there’s a better explanation that we’ve not thought of.
• Maybe the best explanation is not true. (The truth may be surprising!)

Competing Explanations

In IBE we choose a single “winning” explanation from among competing explanations (alternative
hypotheses). Competing explanations are inconsistent: if one is true, the others are false.

In the story, Abby considers two competing explanations.

Hypothesis 4.1: Her cat, Mr. Business, ate the leftover pizza.

Hypothesis 4.2: Her roommate, Zelda, ate the leftover pizza.

Common sense tells us that if her cat ate the pizza then her roommate did not eat it. Of course it’s
possible to imagine a weird story in which they both ate the pizza (e.g. they each ate half of it). Many
of our examples of competing explanations will be like this, very unlikely though perhaps not
impossible that they’re both true.

- 82 -
 The Best Explanation (That We’ve Thought Of!)

It can be difficult to decide which explanation is best. There are many things we could look for in a
good explanation. Here are two important ones.

1) Fit with Background Information

An explanation is better when it fits better with

background information: it’s less surprising,
based on what we already know.

2) Simplicity

A simple explanation is better than a complicated one. One hypothesis is simpler than another if it’s
“easier” to state, if it explains the observation with fewer assumptions or it has fewer parts. As with
fit, usually there is no way to precisely measure simplicity. We normally can’t say that, for example,
one hypothesis is twice as simple (half as complicated) as another. Simplicity is an informal concept.

Example

A very strange reptile called Tanystropheus lived 242 million years ago. Fossils show that about half
the animal’s length was its neck! Why did it have this bizarrely long neck?

(Image: markwitton.com)

Observation 4.3: Tanystropheus had an extremely long neck.

- 83 -
Competing explanations say that its neck was an evolutionary adaptation for different hunting styles.

Hypothesis 4.4: Tanystropheus’ neck was for reaching fish while it stood on land.

(Image: markwitton.com)

Hypothesis 4.5: Tanystropheus’ neck was for sneaking up on fish while it hid in water.

(Image: Emma Finley-Jacob)

Perhaps Hypothesis 4.4 is somewhat simpler. In 2017, it also fit better with background information.
Fossils showed that Tanystropheus’ limbs and tail were not suitably shaped for swimming. Many
paleontologists thought that Hypothesis 4.4 was the best explanation.

But in 2018-20, new fossil studies showed that Tanystropheus

had legs that would be good for jumping forward from an
underwater hiding place and a head with breathing holes on top
of a flat snout+, similar to crocodiles, which hide in water to hunt.
Hypothesis 4.4 fits much worse with all this new information. By
(Image: Emma Finley-Jacob)
2020, Hypothesis 4.5 was the best explanation.


Renesto, S. and Saller, F. (2018) Evidences for a semi-aquatic lifestyle in the Triassic diapsid reptile Tanystropheus. Research
in Paleontology and Stratigraphy 124(1), 23-34.
+
Spiekman, et al. (2020) Aquatic habits and niche partitioning in the extraordinarily long-necked Triassic reptile Tanystropheus.
Current Biology, Aug 6 on-line.

- 84 -
 Inference to the Best Explanation

Main Argument Sub-argument

• EVIDENCE (observation) • HYPOTHESIS fits better with background information

• HYPOTHESIS is the than ALTERNATIVE HYPOTHESES.
best explanation of OR
EVIDENCE. • HYPOTHESIS is simpler than ALTERNATIVE HYPOTHESES.

HYPOTHESIS is the true HYPOTHESIS is the best explanation of EVIDENCE.

explanation of EVIDENCE.

Argument 4.6: [2017] ① Many animals of the Triassic period (252–201 million years
ago) were reptiles. ② One of them, Tanystropheus, had a bizarrely long neck that was
about half of its body length! Probably ③ this was an adaptation for reaching fish while
it stood on land. ④ That seems simpler than the other popular explanation, that a land
animal moved to the water and evolved a long neck to sneak up on fish. ⑤ It also seems
to fit better with what else we know about Tanystropheus from fossils: its body, limbs,
and tail were not well shaped for swimming.

The intermediate conclusion is implicit*: ⑥ “‘Tanystropheus’ neck was an adaptation for reaching fish
while it stood on land’ is the best explanation of its long neck.”, in other words, “③ is the best
explanation of ②.” This argument also has a small additional sub-argument, beyond the extended
pattern, in which background information supports the “fit” claim.

Main Argument Sub-argument

② [Observation] ④
⑥ [H is the best explanation.] ⑤

③ [H is the true explanation.] ⑥

As we saw above, this argument was stronger in 2017 than it is now. Based on the information from
the newer fossil studies, ⑤ and ⑥ now seem to be false; ③ is no longer well supported.

* These examples are written to be as realistic as possible while still conforming to the IBE pattern. Normally some part of
the argument is implicit. People usually either explicitly say that H is the best explanation, and leave it implicit that they think
H is probably true, or they explicitly conclude H, and leave it implicit that they think H is the best explanation.

- 85 -
iii. -CONFIRMATION AND DISCONFIRMATION-

Confirmation

Not every hypothesis is an explanation. In this section, we’ll start to consider hypotheses, such as
hypotheses about the future, that do not explain anything in particular, as well as others that do.

In scientific reasoning, confirmation just means support, not proof. To confirm a hypothesis, we
make and check a prediction, a statement that the hypothesis leads us to expect to be true. To be
useful, the prediction must be something whose truth we can check sooner or more easily or more
directly than the hypothesis. If the prediction is true, this confirms the hypothesis.

A conditional premise connects the hypothesis (antecedent) to the prediction (consequent).

Confirmation

• If HYPOTHESIS then EVIDENCE (prediction).

• EVIDENCE (prediction true)

HYPOTHESIS is true.

Argument 4.7: If mindfulness meditation (sitting quietly,

focusing on your breathing) works then Abby will feel relaxed
after she does it. Later, after meditating, she feels relaxed.
So mindfulness meditation works.

This argument looks like the deductive fallacy AC! But it’s not. Like IBE, confirmation is inductive. It’s
supposed to be strong, not valid.

- 86 -
 Disconfirmation and Background Assumptions

Abby does mindfulness meditation but doesn’t • If HYPOTHESIS then EVIDENCE.

feel relaxed afterwards. This is simple • EVIDENCE is false.
disconfirmation, a regular DC argument:
HYPOTHESIS is false.

We’ll construct a more useful version of this argument that recognizes a background assumption,
a belief we haven’t doubted before now, and which we’ve used to make the prediction. The easiest
way to notice our background assumptions is to think of ways to complete a more complex conditional
sentence: If HYPOTHESIS then EVIDENCE – unless…

Conditional 4.8: If meditation works then Abby will feel

relaxed after meditating – unless Abby drank coffee.

We saw in Unit 3 that unless means “if not“.

Conditional 4.9: If meditation works then Abby will feel

relaxed after meditating – if Abby did not drink coffee.

This shows us the background assumption: Abby did not drink coffee.

Conditional 4.9 has two antecedents, the hypothesis and the background assumption. It says these
are together sufficient for the predicted evidence. The AND operator gives us a way to write this.
Conditional 4.9 is logically equivalent to:

Conditional 4.10: If meditation works and Abby did not drink coffee then she’ll feel
relaxed after meditating.

Conditional 4.10 says: If HYPOTHESIS AND BACKGROUND ASSUMPTION then EVIDENCE.

The prediction is false in our story. So we get a • If (H-AND-BA) then E.

DC argument: • ¬E

¬(H-AND-BA)

- 87 -
 H BA H-AND-BA
1 True True True

In Unit 3, we saw that a conjunction such 2 True False False

as H-AND-BA is false when at least one of 3 False True False

H and BA is false. 4 False False False

This means that: ¬(H-AND-BA) ≡ (¬H)-OR-(¬BA).

This disjunction will be a more useful way of writing the conclusion of our disconfirmation argument
since it directly reminds us that our background assumption might be false instead of the hypothesis.

Disconfirmation (with a Background Assumption)

• If H then E unless ¬BA. / If H-AND-BA then E.

• ¬E

(¬H)-OR-(¬BA)

Argument 4.11: If mindfulness meditation works and Abby didn’t drink coffee then she
will feel relaxed after meditating. But later, after meditating, she doesn’t feel relaxed. So
either mindfulness meditation doesn’t work or she drank coffee.

- 88 -
 -UNIT 4 SKILLS-

You must be able to:

• Think of, or identify in a story, alternative (competing) explanations for an observation.

• Judge the probabilities of hypotheses by their simplicity and fit with background information.

• Reconstruct an IBE argument.

• Recognize hypotheses and predictions and think of background assumptions.

• Construct a disconfirmation argument that includes a background assumption.

- 89 -
 -QUICK TEST QUESTIONS-

1) In an IBE argument…
a) …a prediction explains an observation. c) …a hypothesis explains an observation
b) …an observation explains a hypothesis. d) …a hypothesis explains a prediction.

2) IBE is an inference to the best among explanations that…

a) …are simple. c) …have high probabilities.
b) …have been true in the past. d) …compete.

3) When is an explanation better (more probable)?

a) It is confirmed by observations and it is c) It fits our background information and
supported by predictions. it is confirmed by observations.
b) It is supported by alternative d) It is simple and it fits our background
hypotheses and it is simple. information.

4) A hypothesis can be confirmed only if what?

a) It is simple. c) It is plausible.
b) We can make a prediction from it. d) We can observe it somehow.

5) A confirmation argument has the same pattern as which deductive fallacy?

a) DA c) AC
b) AA d) DC

6) How is a background assumption BA included in the conditional of a disconfirmation argument?

a) If H-AND-BA then E. c) If H then E-AND-BA.
b) IF H then E-OR-BA. d) If H-OR-BA then E.

7) Our prediction is false. What does this tell us?

a) Our hypothesis is false. c) (a) and (b).
b) Some background assumption is false. d) (a) or (b).

- 90 -
 -PRACTICE EXAMS-

PRACTICE EXAM 1

1) Identify the alternative hypotheses (competing with the best explanation). Write the implicit
statement (⑩). Reconstruct the argument in standard form as an extended IBE argument.

① In the Shanghai Pigeon Association’s annual race, pigeons are released far from their

home, and the first to return wins a cash prize for its owner. ② Some pigeons have great

speed and endurance, but ③ this year two returned home in an incredibly short time – the

fastest pigeons ever recorded. ④ The owner claims the breeder bred super-pigeons!

⑤ It’s an outrageous plan but ⑥ the best explanation appears to be that the pigeons’ owner

recaptured them after the race began and took them back to Shanghai on the high speed

train. ⑦ The train makes more sense than a car, with which it would be nearly impossible to

drive to Shanghai fast enough in the dense traffic. ⑧ And the super-pigeon idea, in which a

breeder figured out how to break laws of biology, is way more complicated than the train

scheme. ⑨ It’s amazing the things that people will cheat at.

- 91 -
2) Answer the questions based on the story.

In the 10th century, the Cadronian Kingdom dominated the Eastern Middle region. In the year
978, Princess Yesenia was in line to become the Cadronian queen when her father the king
died. All she had to do to become queen was marry Prince Rastupin. But instead she fled the
castle, giving up her royal status, and lived the rest of her life as a common person.

Historians who study the Cadronian Kingdom have an idea why Yesenia did this: she was
influenced by the ideas of a subversive thinker named Flasgar, who led a movement that
rejected rule by royalty and urged royals to give up their status in solidarity with the common
people. A few of them did. Now the historians are thrilled by the discovery of a scroll that
appears to be a letter Yesenia wrote to her sister in 977, shortly before she fled. They expect
that if their idea is right, the letter will tell her sister her plan to join Flasgar’s movement.

a) What observation do the historians try to explain?

b) What is their hypothesis (explanation)?
c) Think of an alternative hypothesis.
d) What is their prediction?
e) Think of a background assumption of this prediction.

They open the scroll and read the letter. It says nothing about Flasgar’s anti-royalty
movement. Yesenia mostly complains about her royal duties and gossips about other royals.

f) Write a disconfirmation argument that includes the background assumption from (e).

- 92 -
 PRACTICE EXAM 2

1) Identify the alternative hypothesis (competing with the best explanation). Write the implicit
statement (⑧). Reconstruct the argument in standard form as an extended IBE argument.

① The idea of abduction (kidnapping) by aliens has been part of our culture for decades. ②

Most famously, in 1961 Betty and Barney Hill claimed they had been briefly abducted from

their car by a UFO (alien spaceship). But think about it: ③ aliens flying across the galaxy just

to abduct two humans for an hour is a far more complicated explanation of the Hills’ claim

than another theory, that the Hills were delusional (hallucinating or confused).

And although ④ Betty seemed credible at first, ⑤ delusion makes more sense when you

consider that she spent the next twenty years claiming that UFOs were following her, which

even other UFO believers thought was crazy. Moreover ⑥ the fact that no one has ever

produced any physical evidence of alien abduction makes it hard to believe their story. The

conclusion is inescapable: ⑦ the Hills were delusional.

- 93 -
2) Answer the questions based on the story.

Gotham’s train system connects downtown to surrounding areas.

The Ministry of Transportation is reviewing ridership data and
discovers that despite upgrades to the system, ridership has
declined. They think that perhaps people are now using the
ridesharing service Uber instead of the subway.

They want to find out, since an underused train and more street traffic is bad for Gotham. So
they get political support to add an extra fee to Uber rides. The new fee begins. They assume
that if the train ridership decline was because of Uber, next month’s data will show increased
ridership. (The ridership data is based on ticket sales records.)

a) What observation does the Ministry try to explain?

b) What is their hypothesis (explanation)?
c) Think of an alternative hypothesis.
d) What is their prediction?
e) Think of a background assumption of this prediction.

After a month of the new fee on Uber rides, the Transportation Ministry gets new data on train
ridership. Oddly, the data shows no increase in subway ridership.

f) Write a disconfirmation argument that includes the background assumption from (e).

- 94 -
 PRACTICE EXAM 3

1) Identify the alternative hypotheses (competing with the best explanation). Write the implicit
statement (⑪). Reconstruct the argument in standard form as an extended IBE argument.

① In the Czech Republic, archaeologists found 3500-year-old bronze rings of very uniform

size and shape. ② Some archaeologists think that the rings are evidence of an advanced

ancient society: ③ they were a system of money, with a standardized size so that every ring

contained the same amount of valuable bronze. But ④ this explanation is probably wrong. ⑤

The rings were probably just ornaments that all had the same size because they were all cast

from a single mold.

⑥ Though fascinating, the idea that the rings were part of an ancient commercial economy

(people buying things with money instead of trading goods) in which they functioned as a

medium of exchange is needlessly complicated. ⑦ And the ornament explanation makes the

most sense in light of what we already know: ⑧ ancient people made lots of ornaments, and

⑨ permanent commercial economies developed much later in history. Still ⑩ it’s good that

archaeologists are open-minded about ancient peoples, who were more sophisticated than we

often imagine them to have been.

- 95 -
2) Answer the questions based on the story.

Gotham never had a plastics recycling program until last year,

when the city began collecting plastics after a large demonstration
at city hall by the activist group Please No Plastic, who were
protesting against plastic pollution.

PNP is suspicious. They know that plastics recycling programs are mostly ineffective. Recycling
plastic is difficult and expensive, and most companies that use plastic just buy new plastic instead
of recycled. They also know that some companies could lose a lot of revenue if the public stopped
buying their plastic-packaged products. They suspect that the government officials created the
futile recycling program because the companies bribed them to create it in order to make
consumers feel less guilty about buying packaged products.

a) What observation do PNP try to explain?

b) What is their hypothesis (explanation)?
c) Think of an alternative hypothesis.
d) What is their prediction?
e) Think of a background assumption of this prediction.

PNP decide to file a Freedom Of Information request to search the government’s emails for bribe
offers from the companies. They submit all the required FOI forms and after six months receive
the records. They read all the emails and find no bribe offers.

f) Write a disconfirmation argument that includes the background assumption from (e).

- 96 -
 PRACTICE EXAM 4

1) Identify the alternative hypotheses (competing with the best explanation). Write the implicit
statement (⑩). Reconstruct the argument in standard form as an extended IBE argument.

① TV manufacturers used to make 3D TVs that worked with special glasses. But ② the

manufacturers stopped making these TVs. ③ The reason this happened is debated. ④ Maybe

TV stores didn’t promote and demonstrate the 3D technology. ⑤ It’s even possible that

filmmakers who hate 3D organized a campaign to get manufacturers to drop 3D technology out

of respect for traditional film.

⑥ But the best explanation seems to be that TV buyers just found the 3D glasses too annoying.

⑦ This is more sensible considering the fact that the stores had a big profit incentive to sell

the new products, and ⑧ the fact that most TV buyers watch TV with their families in living

rooms where the 3D glasses make everything in the room look blurry. And obviously ⑨ the

glasses explanation is just more straightforward than a campaign by angry filmmakers!

- 97 -
2) Answer the questions based on the story.

Scientists at the National

Organization for the Study of Planets
and Moons (NOSPAM) are studying
Mars, looking for liquid water, which
is very important for life. They are
excited when their telescope shows
wavy patterns on Mars. These could
be a sign of running water. It’s
difficult to be sure, but this is high
enough that it’s worth investigating. (Image: Encyclopædia Britannica)

The European Space Agency will send a space probe to Mars in a few years. The NOSPAM
scientists have an idea. They calculate the amount of water vapour that would be created in
the Mars air by running water. Based on their calculation, they design a new instrument, the
Water Vapour Detector (WAVD), to attach to the probe. They think: “The WAVD will detect
water vapour in the Mars air if those wavy patterns were made by running water.”

a) What observation do the NOSPAM scientists try to explain?

b) What is their hypothesis (explanation)?
c) Think of an alternative hypothesis.
d) What is their prediction?
e) Think of a background assumption of this prediction.

The probe flies to Mars and lands safely. NOSPAM activates the WAVD and it begins measuring
gases in the air. The scientists are disappointed when the WAVD sends its report back to Earth:
no water vapour in the Mars air.

f) Write a disconfirmation argument that includes the background assumption from (e).

- 98 -
 PRACTICE EXAM 5

1) Identify the alternative hypotheses (which compete with the best explanation). Write the implicit
statement (⑧). Reconstruct the argument in standard form as an extended IBE argument.

① Sparrows are little birds known for the males’ birdsong. ② The dominant song across

Canada has had a triplet pattern, while a variant song on the West coast has had a doublet

pattern instead. Probably ③ the doublet pattern is more attractive to female sparrows. That’s

suggested by a surprising recent event: ④ the doublet variant spread across the country,

replacing the triplet pattern.

⑤ One idea, that bird lovers have inadvertently trained the sparrows to sing the doublet

pattern by giving more birdseed to doublet singers, is obviously far too complicated to take

seriously. ⑥ A more reasonable explanation, that the doublet version works better for

establishing territories, is not well supported by other observations: birds singing the doublet

version don’t establish territories more often or quickly. ⑦ Perhaps in the future there’ll be

some new sparrow song!

- 99 -
2) Answer the questions based on the story.

Abby notices that Bob has not said anything about plans for her upcoming 21 st birthday. She
thinks maybe this is because Bob is planning a surprise party for her. This seems like a good
explanation. Then she thinks: “Bob knows how much I love karaoke. If I’m right about his plan,
I bet when I call my favourite karaoke bar, Katie’s Karaoke, they’ll tell me they have a big
reservation on the night of my birthday”.

a) What observation does Abby try to explain?

b) What is Abby’s hypothesis (explanation)?
c) Think of an alternative hypothesis.
d) What is Abby’s prediction?
e) Think of a background assumption of this prediction.

Abby calls Katie’s Karaoke and asks them about reservations on the night of her birthday. They
tell her that they have no reservations on that night.
f) Write a disconfirmation argument that includes the background assumption from (e).

- 100 -
 -ANSWER KEY-

QUICK TEST
1) c 2) d 3) d 4) b 5) c 6) a 7) d

PRACTICE EXAM 1

1) Answer
Alternative hypotheses:
• The owner drove the pigeons back to Shanghai.
• The breeder bred super-pigeons.
Implicit statement: ⑩ The pigeons’ owner recaptured them and took them back to Shanghai on
the high speed train.

Main Argument Sub-argument

③ [Observation] ⑦
⑥ [⑩ is the best explanation of ③.] ⑧

⑩ [Hypothesis] ⑥

Explanation (not required for exam answers)

The alternative hypotheses are not numbered statements in the paragraph. They are mentioned
or referred to in statements that say something about the hypotheses (e.g. that one is simpler).
The role that each statement has in the IBE main argument is given in the answer in [brackets].

- 101 -
2) Answer
a) Princess Yesenia fled the castle and lived the rest of her life as a common person.
b) Yesenia was influenced by the ideas of the subversive thinker, Flasgar.
c) Example answers:
- Yesenia didn’t want to marry Prince Rastupin.
- Yesenia wanted to be free of royal duties.
- Yesenia got a strange brain parasite that made her go crazy.
d) The scroll letter will tell her sister her plan to join Flasgar’s movement.
e) Example answers:
- Yesenia was not worried the letter would be intercepted by royal spies.
- Yesenia trusted her sister not to reveal her plan.
- The letter is not written in a secret code that Yesenia shared with her sister.
f)
• If Yesenia was influenced by Flasgar then the scroll letter will tell her sister her plan to join
Flasgar’s movement unless she was worried the letter would be intercepted by royal spies.
or
If Yesenia was influenced by Flasgar and she was not worried the letter would be intercepted
by royal spies then the scroll letter will tell her sister her plan to join Flasgar’s movement.
• The scroll letter does not tell her sister her plan to join Flasgar’s movement.

Yesenia was not influenced by Flasgar or she was worried the letter would be intercepted by
royal spies.

Explanation
For answer (f), there are two correct patterns for the conditional premise of the argument. Choose
one of them. Either will get full marks on an exam. Both patterns use an example background
assumption, so other answers are correct. Logical operators and conditionals are underlined to
help you read the answer, but you do not need to do this for the exam answer.

- 102 -
 PRACTICE EXAM 2

1) Answer
Alternative hypothesis:
• Aliens flew across the galaxy to abduct two humans for an hour.
Implicit statement: ⑧ Delusion is the best explanation of the Hills’ claims.

Main Argument Sub-argument

③
② [Observation] ⑤
⑧ [⑦ is the best explanation of ②.] ⑥

⑦ [Hypothesis] ⑧

Explanation (not required for exam answers)

The alternative hypotheses are not numbered statements in the paragraph. They are mentioned
or referred to in statements that say something about the hypotheses (e.g. that one is simpler).
The role that each statement has in the IBE main argument is given in the answer in [brackets].

- 103 -
2) Answer
a) Subway ridership has declined despite system upgrades.
b) People are using Uber instead of the subway.
c) Example answers:
- People are riding bicycles.
- People are working from home.
- (Also correct: The computers that count subway riders are broken and missing riders.)
d) Subway ridership will increase after the new Uber fee.
(Also correct: Their data will show an increase in ridership after the fee.)
e) Example answers:
- The new fee is large enough to change people’s travel behaviour.
- People do not switch from Uber to Lyft.
- (Also correct: The computers that count subway riders are not broken.)
f)
• If people are using Uber instead of the subway then subway ridership will increase after
the new Uber fee unless the new fee is not large enough to change people’s travel
behaviour.
or
If people are using Uber instead of the subway and the new fee is large enough to change
people’s travel behaviour then subway ridership will increase after the new Uber fee.
• Subway ridership did not increase after the fee.

People are not using Uber or the new fee was not large enough to change behaviour.

Explanation
The “also correct” answers are not just other example answers, but a different kind of answer.
They concern the information, or the observation itself, rather than the fact observed. For
example, in (c), the first two example hypotheses explain the fact of reduced ridership. The third
hypothesis explains mistaken information about reduced ridership (the ridership is not reduced).

For answer (f), there are two correct patterns for the conditional premise of the argument. Choose
one of them. Either will get full marks on an exam. Both patterns use an example background
assumption, so other answers are correct. Logical operators and conditionals are underlined to
help you read the answer, but you do not need to do this for the exam answer.

- 104 -
 PRACTICE EXAM 3

1) Answer
Alternative hypotheses:
• The rings were an ancient system of money.

Implicit statement ⑩: The best explanation of the rings’ uniform size is they were ornaments
cast from a single mold.

Main Argument Sub-argument

⑥
⑦
① [Observation] ⑧
⑩ [④ is the best explanation of ①.] ⑨

④ [Hypothesis] ⑥

Explanation (not required for exam answers)

The alternative hypothesis is not a numbered statement in the paragraph. It is mentioned or
referred to in statements that say something about the hypotheses (e.g. that one is simpler). The
role that each statement has in the IBE main argument is given in the answer in [brackets].

- 105 -
2) Answer
a) Gotham started a plastics recycling program.
b) Companies that sell products packaged in plastic bribed government officials to create the
program in order to make consumers feel less guilty about buying their products.
c) Example answers:
- The government was genuinely concerned about plastic pollution and thought the program
would do good.
- The government wanted to do something to respond to the protests about plastic.
d) The emails released by the FOI request will contain a bribe offer from the companies.
e) Example answers:
- The companies didn’t bribe the officials with encrypted messages, or in person, or by phone.
- The officials didn’t delete the bribe email.

f)
• If the companies bribed the government officials and they officials didn’t delete the bribe
email then the emails released by the FOI request will contain a bribe offer from the
companies.
• The emails released by the FOI request do not contain a bribe offer from the companies.

The companies did not bribe the officials or the officials did delete the bribe email.

Explanation
For answer (f), logical operators and conditionals are underlined to help you read the answer, but
you do not need to do this for the exam answer.

- 106 -
 PRACTICE EXAM 4

1) Answer
Alternative hypotheses:
• TV stores didn’t promote and demonstrate 3D technology.
• Filmmakers who hate 3D coordinated a successful campaign to pressure manufacturers into
dropping the technology out of respect for traditional film.
Implicit statement: ⑩ TV buyers found the 3D glasses too annoying.

Main Argument Sub-argument

⑦
② [Observation] ⑧
⑥ [⑩ is the best explanation of ②.] ⑨

⑩ [Hypothesis] ⑥

Explanation (not required for exam answers)

The alternative hypotheses are not numbered statements in the paragraph. They are mentioned
or referred to in statements that say something about the hypotheses (e.g. that one is simpler).
The role that each statement has in the IBE main argument is given in the answer in [brackets].

- 107 -
2) Answer
d) There are wavy patterns on Mars.
(Also correct: Images from telescopes show wavy patterns on Mars.)
e) There is running water on Mars.
f) Example answers:
- Wind made the wavy patterns.
- Aliens drew the wavy patterns as a giant art project.
- (Also correct: There is a wavy shaped smudge on the lens of the telescope.)
g) There is water vapour in the Mars air.
(Also correct: The WAVD will detect water vapour in the Mars air.)
h) Example answers:
- The water vapour has not escaped into space.
- (Also correct: NOSPAM’s calculations were correct.)
- (Also correct: The WAVD works as designed (is not damaged in its flight to Mars).)

i)
• If there is running water on Mars then there is water vapour in the Mars air unless the
water vapour has escaped into space.
or
If there is running water on Mars and the water vapour has not escaped into space then
there is water vapour in the Mars air.
• There is no water vapour in the Mars air.

There is not running water on Mars or the water vapour has escaped into space.

Explanation
The “also correct” answers are not just other example answers, but a different kind of answer.
They concern the information, or the observation itself, rather than the fact observed. For
example, in (c), the first two example hypotheses explain the fact of wavy lines on Mars. The
third hypothesis explains mistaken information about wavy lines (there are no wavy lines).

For answer (f), there are two correct patterns for the conditional premise of the argument. Choose
one of them. Either will get full marks on an exam. Both patterns use an example background
assumption, so other answers are correct. Logical operators and conditionals are underlined to
help you read the answer, but you do not need to do this for the exam answer.

- 108 -
 PRACTICE EXAM 5

1) Answer
Alternative hypotheses:
• Bird lovers have trained sparrows to sing the doublet pattern by giving more birdseed to
doublet singers.
• The doublet version works better for establishing territories.
Implicit statement: ⑧ The greater attractiveness of the doublet pattern to females is the best
explanation of it spreading across the country.

Main Argument Sub-argument

④ [Observation] ⑤
⑧ [③ is the best explanation of ④.] ⑥

③ [Hypothesis] ⑧

Explanation (not required for exam answers)

The alternative hypotheses are not numbered statements in the paragraph. They are mentioned
or referred to in statements that say something about the hypotheses (e.g. that one is simpler).
The role that each statement has in the IBE main argument is given in the answer in [brackets].

- 109 -
2) Answer
a) Bob has said nothing about her upcoming birthday.
b) Bob is planning a surprise birthday party for Abby.
c) Example answers:
- Bob can’t afford to buy her a present and is hoping that she will forget her own birthday.
- Bob is so busy studying philosophy that he forgot about her birthday.
d) Katie’s Karaoke will tell Abby that there is a big reservation on the night of her birthday.
e) Example answers:
- Bob is not planning a surprise party somewhere else instead.
- Bob has not instructed Katie’s Karaoke to lie to Abby.
f)
• If Bob is planning a surprise party for Abby then Katie’s Karaoke will tell Abby that there is
a big reservation on that night unless Bob instructed Katie’s to lie to Abby.
or
If Bob is planning a surprise party for Abby and he has not instructed Katie’s Karaoke to lie
to Abby then Katie’s will tell Abby that there is a big reservation on that night.
• Katie’s does not tell Abby that there is a big reservation on that night.

Bob is not planning a surprise party for Abby or he has told Katie’s to lie to Abby.

Explanation
For answer (f), there are two correct patterns for the conditional premise of the argument. Choose
one of them. Either will get full marks on an exam. Both patterns use an example background
assumption, so other answers are correct. Logical operators and conditionals are underlined to
help you read the answer, but you do not need to do this for the exam answer.

- 110 -
 ~5~
-CONFIRMATION BIAS-

Confirmation bias is a bad use of evidence that makes a hypothesis seem better supported than it
really is. This bias is a natural human tendency. We all instinctively look for evidence that supports
the beliefs we already have. We might do this when we want our belief to be true (wishful thinking),
but we do it even with beliefs we wish were false. While it’s very difficult to suppress confirmation
bias, we can try to counter it by remembering to look for evidence that we’re wrong.

Here are four ways confirmation bias can happen.

1) Ignoring Alternative Explanations (of confirming evidence)

An observation supports a hypothesis when the hypothesis explains it. If there are other explanations,
the observation is weaker evidence for the hypothesis, since we can explain the observation without
believing the hypothesis. So one form of confirmation bias is ignoring alternative explanations.

Confirmation Bias 5.1: Bob is walking around his city’s harbour. Past the nice part of the
harbour, near an old unused dock, there’s a broken boat stuck in the sand. Bob has read
stories about sea monsters reported throughout history. He thinks: “I bet this boat was
broken by a sea monster! A sea monster would probably crash into boats and leave them
broken like this.”

When Bob finds evidence (the broken boat) that’s explained by a sea monster, he constructs a
confirmation argument that supports that hypothesis. But there are alternative – and better! –
explanations. The boat is near an old dock where there’s little activity. The boat could have been
smashed in a storm or by a larger boat, or just rotted without maintenance, etc. By ignoring these
explanations, Bob makes the boat seem to be stronger support for the sea monster than it really is.

- 111 -
 2) Biased Search for Evidence

Confirmation bias can occur in the way that someone looks for, collects, or remembers evidence. They
select evidence that confirms their hypothesis and ignore, or filter out, evidence that disconfirms it.
We should judge the hypothesis based on our total evidence, and we should actively search for
disconfirming evidence to counter our natural inclination to search for confirming evidence.

Confirmation Bias 5.2: “If there’s a sea monster,” Bob thinks,

“probably it eats a lot of fish and people catch fewer fish here
than normal.” He asks fishing people about their catches. He talks
to some in the harbour who tell him they catch a normal amount.
He asks others returning on a fishing boat and they caught fish
just as easily as they do elsewhere. He asks a man fishing north
of the harbour who tells him that fishing is a waste of time there
– he catches very few fish. “Aha”, Bob thinks, “The sea monster
has been hungry!”

Bob uses the small fish catch of one person to confirm his sea monster hypothesis but ignores other
people’s normal fish catches. He selects evidence to support the hypothesis he already believes.

3) Ad Hoc Explanation (of disconfirming evidence)

If disconfirming evidence is too obvious to ignore, someone with confirmation bias might give an ad
hoc explanation instead. This is an excuse someone makes up to “protect” their hypothesis from
disconfirmation. They have evidence that should make them think their hypothesis might be false, but
they’re convinced it’s true, so they think of some other way to explain the disconfirming evidence.

Confirmation Bias 5.3: Bob writes letters to the local

news demanding that the city government search for the
sea monster. The news publishes an article. The city puts
an underwater camera in the harbour for a few weeks.
When they announce that the camera found no sea
monster, Bob thinks: “Probably the government doesn’t
want citizens to panic and flee the city. They’re keeping
the sea monster a secret.”

- 112 -
Bob would accept an announcement that the city had a photo of a sea monster as confirmation of his
hypothesis, so he should accept the announcement of no photo as disconfirmation. But he doesn’t.
He’s still sure that there’s a sea monster. To protect his hypothesis from disconfirmation, he explains
the announcement by saying that the city doesn’t want people to panic, though he has no good reason
to think that.

4) Imprecise Prediction

[This error is not usually called confirmation bias. However the effect of making the error is that
confirmation becomes “too easy” and people may overestimate how well a hypothesis is supported,
so we’ll include it as a form of confirmation bias.]

An imprecise prediction is not specific enough for strong confirmation. The less precise it is, the
more possible ways it could be true. But in many of these ways, the hypothesis will be false. A
prediction is supposed to be a test of the hypothesis, but this test is too easy! Passing it shows almost
nothing. An imprecise prediction is another way of protecting a hypothesis from disconfirmation.

Confirmation Bias 5.4: “A sea monster is a big creature that

would cause some disturbance,” Bob thinks. He spends the week
looking around and talking to people. At a cargo loading dock,
workers tell him that a container full of TVs somehow fell into the
water last month. “Wow, that’s certainly a disturbance,” Bob
thinks. “This harbour has a sea monster!”

Bob’s prediction is “some disturbance” around the harbour. Many different things could count as a
disturbance. This prediction is not very precise (specific). So it’s not surprising, and not strongly
confirming, that the prediction is true. There would likely be a disturbance of some sort even if there’s
no sea monster. It’s too easy for the sea monster hypothesis to pass this test.

- 113 -
 -UNIT 5 SKILLS-

You must be able to:

• Recognize and describe examples of confirmation bias in a story.

- 114 -
 -QUICK TEST QUESTIONS-

1) We’ll overestimate the support for our hypothesis by ignoring alternative explanations of what?
a) Confirming evidence. c) Our prediction.
b) Disconfirming evidence. d) Our test.

2) Which is not an example of a biased search for evidence?

a) Ignoring false predictions. c) Remembering only supporting evidence.
b) Looking only for confirmation. d) Testing only true hypotheses.

3) What is explained by an ad hoc explanation?

a) Confirming evidence. c) A true hypothesis.
b) Disconfirming evidence. d) A false hypothesis.

4) An imprecise prediction is bad because it’s more likely to be _____ when the hypothesis is _____.
a) …true… true c) …false… true
b) …true… false d) …false… false

- 115 -
 -PRACTICE EXAMS-

PRACTICE EXAM 1

Identify the hypothesis. Then identify two examples of confirmation bias in the story, referring to
specific details and using specific concepts from this unit.

The war between Azmakia and its neighbour Blastamak has been
raging for months. Many people living in northern Azmakia have
cultural and linguistic ties to Blastamak, and Blastamak has
invaded the region to “liberate” them.

The president of Blastamak gets daily briefings from his military generals as he eagerly waits for the
victory that he’s sure Blastamak will achieve. For a while, the generals report normal updates. When
the army captures a town, the president is happy and promotes the general who reports this. But
when another general tells him that a whole army division was destroyed by an Azmakian counter-
offensive, the president flies into a rage and demotes the general to dish-washing duty in the
cafeteria. The other generals are now careful about what news they tell the president.

The president gives a speech to the citizens, who are now tired of the war. He assures them that
they will finally win the war after another 20,000 Blastamakans join the army. Some people in the
audience wave a banner: WE WON’T FIGHT YOUR UNJUST WAR. Security guards immediately arrest
the protestors. The president pauses his speech and looks angrily at them. He says to the crowd,
who are watching the security guards take the protestors away: “Those are just some scumbag
foreign agents who snuck into our country to spread dissent and weaken us. But they won’t succeed.
Blastamak will be victorious!”

- 116 -
 PRACTICE EXAM 2

Identify the hypothesis. Then identify two examples of confirmation bias in the story, referring to
specific details and using specific concepts from this unit.

Bob is buying a hamster. The pet store owner points to one.

“That’s a Mexican hopping hamster. These evolved to be very
active because they’re hunted by snakes.” Bob buys it and
names her Kalypso. After several days of chatting to her, he
gets a weird feeling. “I think that Kalypso understands me!”

“This is amazing”, he thinks. “I wonder why the pet store owner never noticed this. It’ll be great
to have a new friend. I hope she stays interested in my life.” One day he says “You know, Kalypso,
I work really hard at my job and I deserve a raise. Do you think I should ask for one?” She hops
up and down in her cage. “She thinks I should! What a nice and clever little furry friend I have!”

The next day after feeding her, he decides to test Kalypso to make sure that he’s right about her
abilities. He puts three blocks in her cage: a square, a circle, and a triangle. He says: “OK, Kalypso,
show me how smart you are. Go sit on the square block.” She wanders over to the triangle and
grooms her fur. “Huh – I bet she’s probably extra hungry today and not in the mood for tests.”

He’s now convinced and starts posting about Kalypso on Snapagram for his friends to read.

- 117 -
 PRACTICE EXAM 3

Identify the hypothesis. Then identify two examples of confirmation bias in the story, referring to
specific details and using specific concepts from this unit.

Abby is alarmed about her town’s new mayor, Ladnod Prumt. She
suspects that Prumt is corrupt (that he uses his political power for
his own personal gain). She wants the media to expose him as
corrupt so that he’s removed from office.
(Image: Vecteezy.com)

Abby figures that if Prumt really is making secret illegal deals (money or gifts in exchange for his
votes on city council), he’ll have to meet people. She follows him for a few weeks, watching for
meetings. One day he meets another man in a coffee shop, and they talk for half an hour.
“Gotcha”, Abby thinks. “I am going to expose your corrupt ways, Ladnod.” She takes a photo.

Of course the main thing to look for in a corrupt politician is fancy expensive possessions or
activities. She checks Prumt’s house. It’s pretty normal and needs some new paint. He drives a
six year old Honda Civic. Later she spies on him shopping for a new suit at Suits4Less. One day
she hides in a bush while he’s getting his mail and spies an Omega watch on his wrist. She Googles
the price: $7,500.

Abby starts writing emails to the media to alert them that Prumt is corrupt.

- 118 -
 PRACTICE EXAM 4

Identify the hypothesis. Then identify two examples of confirmation bias in the story, referring to
specific details and using specific concepts from this unit.

Venty Capital is a venture capital firm looking to invest in an

innovative startup company. They’re interested in
Wackadoodle Industries. Venty thinks that Wackadoodle’s
bold, outside-the-box thinking on how to streamline product
development may shake up the industry and make
Wackadoodle the next Amazon or Tesla.

One thing Venty looks at in a startup is whether it’s building relationships with potential customers
who are experts about the industry. That could indicate that Wackadoodle is a good investment.
So Venty is excited to learn that Wackadoodle is in talks with Lambast Technologies, whose chief
operations officer is the Wackadoodle founder’s uncle. Lambast gets government contracts and
awards many sub-contracts. “Lambast would recognize a solid company!” Venty thinks.

Venty feels good about Wackadoodle, but still wants a tour of their headquarters. When they
arrive, some of Wackadoodle’s employees have just put out a fire they accidentally started in the
workshop. Many offices are empty and Venty eventually finds everyone crowded into one office
playing a PlayStation 5. Down the hall, two engineers are drawing diagrams on a whiteboard.
“Hey, look at that,” Venty says. “These Wackadoodle folks seem to be real pros!”

They arrange to transfer $10 million to Wackadoodle.

- 119 -
 PRACTICE EXAM 5

Identify the hypothesis. Then identify two examples of confirmation bias in the story, referring to
specific details and using specific concepts from this unit.

Vaping is popular in Gotham and thousands of people have

lung illnesses. The main problem is sarsaparilla flavour vape
juice, which is dangerous and addictive, so Gotham banned
it. Prohibition led to a black market. The Eastside Sarsaparilla
Gang sells it illegally on the street.

The ESG is afraid the Gotham Police Department is watching them. They meet at their
headquarters and the leader says: “Hey everyone, if the GPD is watching us, there will probably
be something unusual around here. Let’s look.” They walk around the block and notice a parked
van that isn’t normally there. They start to panic. “The GPD is going to arrest us!”

The ESG is making deliveries and one says: “There’s a helicopter flying around. Let’s get a look
at it. I bet we’ll see GPD markings. Then we’ll know for sure we’re being watched.” They go to a
bridge to look at the helicopter with binoculars. On the side it says NEWS 1120 Traffic Chopper.
“Wow, the GPD is clever! They’ve disguised their helicopter to look like it’s for a news station!”

They look at each other and don’t need to say anything else. They race back to the headquarters,
grab the cash, and get the hell out of Gotham.

- 120 -
 -ANSWER KEY-

QUICK TEST
1) a 2) d 3) b 4) b

PRACTICE EXAM 1

Answer
Hypothesis: Blastamak will win the war against Azmakia.

The president shows confirmation bias in supporting his hypothesis.

1) Biased search for evidence. The generals try to report the army’s victories and losses to the
president, but the president wants to hear only about victories. The generals learn they must
report only news that the president wants to hear. So the president will come to believe that the
Blastamakan army is succeeding more than it really is.
2) Ad hoc explanation (of disconfirming evidence). To win the war, Blastamak needs more people to
join the army. The president gets evidence (the protest) that the remaining citizens are not willing
to join. But instead of questioning whether victory is possible, the president says the protesters
are foreign agents, even though there is no particular reason to think this. It looks like he’s
protecting his hypothesis from disconfirmation.

Alternative answers (worth part marks)

1) Ignoring and alternative explanation. The obviously alternative explanation for the banner is that
it reflects the citizens’ genuine discontent and unwillingness to fight.

Explanation
Most of the marks for this type of question are for applying the concept of a specific form of
confirmation bias (one of the four in this unit) to specific details in the story, which may be quoted.
For full marks, you should briefly describe what’s happening in a way that shows that you understand
how someone is gathering evidence in a way that is likely to make them more confident than they
should be.

- 121 -
 PRACTICE EXAM 2

Answer
Hypothesis: Kalypso understands Bob.

Bob shows confirmation bias in supporting his hypothesis.

3) Ignoring alternative explanations of confirming evidence. Kalypso’s hopping seems to be
confirming evidence, but she is a Mexican hopping hamster. Maybe she jumps around when Bob
asks about the raise because she’s naturally hoppy, not because she understands him.
4) Ad hoc explanation (of disconfirming evidence). When Kalypso fails to show that she understands
a simple command, Bob explains his false prediction by thinking that she’s extra hungry and not
in the mood. But he had no other reason to think this. It looks like an excuse he gives to protect
his hypothesis from disconfirmation.

Alternative answers (worth part marks)

2) Imprecise prediction. Although it’s not explicit, Bob seems to have predicted that Kalypso will
show excitement when he asks her a question. But this prediction is imprecise. Many different
hamster actions (hopping, running, quivering, squeaking) might count as showing excitement, so
this prediction had a high chance of being true even if Kalypso doesn’t understand Bob.
3) Biased search for evidence. Although Bob doesn’t ignore the disconfirming evidence (he explains
it), he also doesn’t seem to be seriously interested in disconfirming his hypothesis. Once he gets
the idea that Kalypso understands him, he is interested only in evidence that supports this.

- 122 -
 PRACTICE EXAM 3

Answer
Hypothesis: Prumt is corrupt.

Abby shows confirmation bias in supporting her hypothesis.

1) Imprecise prediction: “Meet people” is not very precise. Meeting anyone, anywhere, anytime?
There are so many ways this could happen, it’s very likely to be true even if Prumt is not corrupt.
2) Biased search for evidence: She seems to ignore or discount the disconfirming evidence that
Prumt’s house, car, and clothes are not fancy or expensive. Instead she selects and focuses on
the confirming evidence of his high-end wristwatch.

- 123 -
 PRACTICE EXAM 4

Answer
Hypothesis: Wackadoodle is a good startup company to invest in.

Venty shows confirmation bias in supporting their hypothesis.

1) Ignoring an alternative explanation of confirming evidence. Lambast Industries’ chief operations
officer is the Wackadoodle founder’s uncle. Maybe Lambast’s interest in Wackadoodle is because
of nepotism or family loyalty, not because Wackadoodle is a good company.
2) Biased search for evidence. Venty seems to ignore or discount the signs of incompetence (starting
fires) and bad work ethic (playing video games) at Wackadoodle. Instead they select and focus
on the confirming evidence of the engineers drawing diagrams.

- 124 -
 PRACTICE EXAM 5

Answer
Hypothesis: The GPD is watching the ESG.

The ESG shows confirmation bias in supporting their hypothesis.

1) Imprecise prediction. “Something unusual” is not very precise. Anything at all unusual? An object?
An event? There are so many ways for something unusual to happen, this is very likely even if
the GPD is not watching them.
2) Ad hoc explanation of disconfirming evidence. They predict GPD markings on the helicopter, but
when there are none, they explain the false prediction by thinking that the GPD disguised the
helicopter. Instead of lowering their confidence in their hypothesis, they make an excuse to
protect it from disconfirmation and then are even more convinced.

Alternative Answers (worth part marks)

1) Ignoring an alternative explanation of confirming evidence. Although there are no specific details
in the story that suggest an alternative explanation, the confirming evidence (the parked van) is
a very ordinary thing. It is easy to think of many explanations other than the ESG’s explanation,
e.g. the van belongs to a contractor doing repair work for someone.
2) Biased search for evidence. Although the ESG doesn’t ignore or dismiss the disconfirming evidence
(they explain it), they also don’t seem to be seriously interested in disconfirming their hypothesis.
Once they get the idea that the GPD is onto them, it seems that nothing will reassure them.

- 125 -
 ~6~
-PROBABILITY-

i. -PROBABILITY-

The Concept of Probability

A probability is a number from 0 to 1 that is assigned to an event or statement. It expresses the

likelihood that the event happens or that the statement is true.

1 100% The extremes of 1 (100%) (certainly happens) and 0 (0%)

(certainly does not happen) apply in special cases, but not in most
0.5 50% real-world situations where it’s possible (a non-zero chance) for
some very strange things to happen. A probability of 50% says
0 0% that an event is equally likely to happen and to not happen.

Sometimes a probability can be directly calculated based on the

details of a situation. For example, we flip a coin. There are two
possible outcomes: Heads and Tails. Since Heads is one of two
possible outcomes, the probability of Heads – P(Heads) – is 1/2.

Basic Probability Calculation (for equally likely outcomes)

A-Outcomes
P(A) =
Total Possible Outcomes

This calculation works with a fair coin that is equally likely to land Heads and Tails and can do nothing
else (e.g. land on its edge, explode in the air, etc.). This gives the same fraction that we’d get if we
flipped the coin again and again, counting the total flips and the Heads flips: 1/2. That probability
would be based on the frequency of the Heads event, how often Heads happens when flipping a coin.

Unit 6 examples use these sorts of probabilities. In Unit 7 we’ll also see examples of another sort.

- 126 -
 Chance and Odds

Chance is often just another word for probability. As we saw above, we can express the chance of A
as a fraction: [A outcomes]-in-[Total outcomes]

Odds is a different way to give a probability.

Odds is like the score in a sports match:
[A outcomes]-to-[B outcomes] or
[A outcomes]-to-[Non-A outcomes].

A outcomes + Non-A outcomes = Total outcomes. So we can convert odds into chance.

Odds of A Chance of A
[A outcomes]-to-[Non-A outcomes] [A outcomes]-in-[Total outcomes]

2:1 2
/3
=
2-to-1 2-in-3 (2-in-(2+1))

Conditional Probability and Independence

A conditional probability P(A|B) is the chance that A happens given (assuming) that B happens.
When B happens, does that change (raise or lower) the chance of A?

Independent Not Independent

P(A) = P(A|B) P(A) ≠ P(A|B)
B does not change the chance of A. B changes the chance of A.

Coin flips are independent. Imagine flipping a coin nine times and getting nine Heads.

?

Vocabulary: Sometimes people say “odds” when they mean chance (e.g. “the odds are 1-in-100”) or “chance” when they
mean odds (e.g. “there’s a 50-50 chance”).

- 127 -
Does it feel that Heads is less likely (Tails is more likely) now? If you’re like many people, it does. But
the previous nine flips don’t matter. P(Heads|9Heads) = P(Heads) = 1/2. The gambler’s fallacy is
the mistaken belief that an event is less likely when events like it happened recently, or more likely
when events like it did not happen recently, though in fact the event is independent of its history.

Some events are not independent.

Surf: Abby surfs.
SharkBite: Abby is bitten by a shark.

P(SharkBite|Surf) ≠ P(SharkBite). Shark bites are more common

among surfers than among people generally (everyone). If we
know that Abby surfs, it’s more likely that she’s bitten by a shark:
P(SharkBite|Surf) > P(SharkBite).

Everyone Shark Bite

Victims
Surfers

Shark bites are extremely rare among everyone: P(SharkBite) = extremely low.

Everyone Shark Bite

Victims

- 128 -
Shark bites are more common (although still rare) among surfers. The chance of SharkBite, given that
Abby surfs, is low: P(SharkBite|Surf) = low.

Surfers
Shark Bite
Victims

Suppose we know that Abby was bitten by a shark. What’s the chance that she surfs? This is a
completely different conditional probability. Most shark bite victims are surfers! The chance that Abby
surfs, given that she was bitten by a shark, is high. P(Surf|SharkBite) = high.

Shark Bite
Victims Surfers

- 129 -
ii. -DISJUNCTION-

Disjunction Rule (Special Case: Mutually Exclusive Events)

Twenty UFOs invade Earth: 5 from planet Xenu, 8 from Zargon, and 7 from other planets.

Xenu Zargon Other Planet

(5) (8) (7)

Disjunction uses an addition rule. P(Xen-OR-Zar) = P(Xen) + P(Zar) = 5/20 + 8/20 = 13

/20.
This simple rule works here because this is a special case where the events (possibilities) are mutually
exclusive: a UFO cannot be from more than one planet.

Disjunction Rule

Six UFOs are flown by aliens who plan to abduct (kidnap) humans to do weird experiments on them.

The chance that a UFO is from Zargon OR it’s Abducting cannot be P(Zar) + P(Abd) = 8/20 + 6/20. That
addition double counts 3 Abducting Zargon UFOs.

Abducting Non-abducting
(6) (14)

- 130 -
To correct this, we subtract the joint probability P(Zar-AND-Abd), the chance of both events
happening. P(Zar) + P(Abd) – P(Zar-AND-Abd) = 8/20 + 6/20 – 3/20 = 11
/20.

Disjunction Rule

P(A-OR-B) = P(A) + P(B) – P(A-AND-B)

In the previous example, the special case of the disjunction of the mutually exclusive events, the
subtraction doesn’t matter since P(Xen-AND-Zar) = 0.

Event A either happens or it doesn’t happen. One of A and ¬A must be true. So P(A-OR-(¬A)) = 1.

The disjunction rule tells us that P(A-OR-(¬A)) = P(A) + P(¬A) – P(A-AND-(¬A)) = 1.

Since A and ¬A are mutually exclusive, P(A-AND-(¬A)) = 0. Therefore: P(A) + P(¬A) – 0 = 1.

Therefore P(A) + P(¬A) = 1. Therefore:

Negation Rule

P(A) = 1 – P(¬A)
P(¬A) = 1 – P(A)

- 131 -
iii. -CONJUNCTION-

Conjunction Rule (Special Case: Independent Events)

We just saw that to calculate the chance of a disjunction, we sometimes need to know the chance of
a conjunction, a joint probability. We flip two fair coins, A and B. What’s the chance that both coins
land Heads? Counting possible outcomes, we see the answer: 1-in-4.

{AHeads, BHeads} {AHeads, BTails} {ATails, BHeads} {ATails, BTails}

Conjunction uses a multiplication rule. P(AHeads-AND-BHeads) = P(AHeads) x P(BHeads) = 1/2 x 1/2

= 1/4. This simple rule works here because this is a special case where the events are independent.

Conjunction Rule

What is P(Zar-AND-Abd)?

P(Zar) = 8/20 and P(Abd) = 6/20. Multiplying these gives us: 8/20 × 6/20 = 3/25. This is wrong. As we
saw in the previous section, there are:
3-in-20 Abducting Zargon UFOs

Xenu Zargon Other Planet Abducting Non-abducting

(5) (8) (7) (6) (14)

- 132 -
The problem is that Zargon and Abducting are not independent. We must multiply with:

P(Abd|Zar) to show how Zargon changes P(Zar|Abd) to show how Abducting

the chance of Abducting. changes the chance of Zargon.

P(Abd|Zar) × P(Zar) P(Zar|Abd) × P(Abd)

3 8 3 3 6 3
/8 × /20 = /20 /6 × /20 = /20

Conjunction Rule

P(A|B) × P(B) = P(A-AND-B) = P(B|A) × P(A)

Disjunction of Conjunctions

We can combine the conjunction and disjunction rules.

A UFO lands and aliens come Xenu Zargon Other Planet

out. What’s the chance they AND OR AND OR AND
plan to abduct us? There are Abducting Abducting Abducting
three ways this could be an
abducting UFO:

The conjunction rule gives us the chance of each possibility. Since these are mutually exclusive (a
UFO cannot be from more than one planet), we add them according to the disjunction rule. The total
or overall probability P(Abd) is the sum P(Xen-AND-Abd) + P(Zar-AND-Abd) + P(OP-AND-Abd).

P(Abd) = (P(Abd|Xen) × P(Xen)) + (P(Abd|Zar) × P(Zar)) + (P(Abd|OP) × P(OP))

= (1/5 × 5/20) + (3/8 × 8/20) + (2/7 × 7/20)
1 3 2
= /20 + /20 + /20
6
= /20

We can see in the picture that this answer is correct: 6-in-20 UFOs are abducting.

- 133 -
 Three Common Confusions

1) Is the | symbol in P(A|B) a mathematical instruction?

No! (A|B) means “A given (assuming) B”. The | symbol does not mean “divided by”, “plus”, or
“multiplied by”. A and B are events (or statements), not numbers!

2) Are joint probability and conditional probability the same?

No! They are completely different. We use a conditional probability to calculate a joint probability.

Joint Probability Conditional Probability

P(A-AND-B) P(A|B)
Chance of two events: A and B Chance of one event: A

3) Are mutual exclusion and non-independence the same?

No! Mutual exclusion is a special case of non-independence. Non-independent means that when
one event happens, the chance of the other event changes. Mutually exclusive means that when
one event happens, the chance of the other event changes to 0.

- 134 -
iv. -EXPECTED VALUE-

Sometimes when deciding what to do, a rational rule is: maximize expected value. We must know the
probability and the value (in some units) of each possible outcome of the actions we are considering.
An action’s expected value is the probability-weighted average of the values of its possible
outcomes. Like a disjunction of conjunctions, this is a “sum of products” calculation.

Expected Value (of Action with Possible Outcomes A and B)

Expected value = (A-Value × P(A)) + (B-Value × P(B))

Bob gets a chicken. He wants the most eggs per

week. What colour should he paint her coop? He
reads about how chickens respond to different
colours. In a blue coop, there’s a 60% chance his
chicken will be happy and lay 5 eggs/week, but a
40% chance she’ll be unhappy and lay 1
egg/week. With green, there’s an 80% chance of
3 eggs/week and a 20% chance of 2 eggs/week. (Image: Big_Ryan / Getty Images)

Assuming that maximum eggs/week is all that matters, each possible outcome of each action (colour)
has a known value (in units of eggs/week). And it has a probability. So Bob can calculate an expected
value for each colour and choose the colour that maximizes expected value.

Blue: (5 eggs/wk × 0.6) + (1 egg/wk × 0.4) = 3.4 eggs/wk

Green: (3 eggs/wk × 0.8) + (2 eggs/wk × 0.2) = 2.8 eggs/wk

Bob should paint the coop blue to maximize the expected eggs/wk.


This is not realistic! Maybe Bob is happy with 4 eggs/week and a 5 th egg has no value to him. Maybe he needs at least 2
eggs/week and 1 egg would be a disaster. Maybe the chicken’s happiness has value that should be counted (a moral
consideration). Maximizing expected value in real life is complicated.

- 135 -
v. -AT LEAST ONE-

Chance of At Least One (Equally Probable Cases)

We roll some dice. What’s the chance of rolling any 3s (at least one 3)? Since any means “not none”,
the chance of any equals 1 minus the chance of none. With more and more dice, the chance of rolling
no 3s gets closer and closer to 0 and the chance of rolling any 3s gets closer and closer to 1.

P(At-Least-1 A) (with N equally probable cases)

P(AL1-A(N Cases)) = 1 – P(¬A)N = 1 – (1 – P(A))N

For P(AL1-Roll3(N Dice)) we put an exponent N on P(¬Roll3) for the number of dice. That tells us to
multiply the chance of ¬Roll3 by itself for every dice we roll. With one dice, the exponent is 1, and
the chance of rolling at least one 3 (AL1Roll3(1Dice)) is simply the chance of rolling 3: 1/6.

P(AL1-Roll3(1Dice)) = 1 – P(¬Roll3)1 = 1 – (1 – 1/6)1 = 1– 5

/6 ≈ 0.167

The solution is the same for more than one dice. We just adjust the exponent. For 2 dice:

P(AL1-Roll3(2Dice)) = 1 – P(¬Roll3)2 = 1 – (1 – 1/6)2 = 1– 25

/36 ≈ 0.306

The solution works the same for any number of cases. For 20 dice, the numbers in the fraction are
large, but the structure of the solution is the same.

P(AL1-Roll3(20Dice)) = 1 – (1 – 1/6)20 = 1– 95,367,431,640,625

/3,656,158,440,062,976 ≈ 0.974

Since the chance of rolling no 3s never reaches 0, the chance of at least one 3 never reaches 1.

- 136 -
 -UNIT 6 SKILLS-

You must be able to:

• Recognize and write probabilities, including conditional probabilities in the chance and odds forms.

• Distinguish independent and non-independent events; recognize the gambler’s fallacy.

• Use the disjunction and conjunction rules separately and together (disjunction of conjunctions).

• Calculate the chance that an event happens at least once.

• Determine a correct decision based on expected value.

- 137 -
 -QUICK TEST QUESTIONS-

1) “More likely to happen than not to happen” means what probability?

a) 100% c) ≥ 50%
b) > 50% d) > 0%

2) Convert “5-to-3 odds” into a chance.

a) 5-in-3 c) 2-in-3
b) 3-in-5 d) 5-in-8

3) What does P(A|B) mean?

a) Chance of A times chance of B. c) Chance of A assuming B.
b) Chance of A divided by chance of B. d) Chance of A and chance of B.

4) Do P(A-AND-B) and P(A|B) mean the same thing?

a) Yes. c) No.
b) No, but close enough that you don’t need d) NO!!!
to worry about the difference.

5) Which is true?
a) If A and B are mutually exclusive, they c) If A and B are not independent, they are
are independent. mutually exclusive.
b) If A and B are mutually exclusive, they d) A and B are mutually exclusive if and
are not independent. only if they are not independent.

6) Which says “the probability that A happens given that B happens”?

a) P(A|B) c) P(A)|P(B)
b) P(A)|P(B) d) P(B|A)

7) When are two events A and B independent?

a) P(A) = P(B|A) c) P(A) ≠ P(A|B)
b) P(A) ≠ P(B|A) d) P(A) = P(A|B)

8) In the gambler’s fallacy, someone mistakenly thinks the chance of an event depends on ________.
a) nothing c) its history
b) its future d) a bet that they have made

- 138 -
9) “Few A are B.” What probability does this give us?
a) P(B|A) is low. c) P(A|B) is high.
b) P(A|B) is low. d) (a) and (b).

10) When are two events mutually exclusive?

a) They cannot both happen. c) Neither of them can happen.
b) They can both happen. d) If one happens, the other must happen.

11) What is P(A-OR-B)?

a) Between P(A) and P(B). c) ≥ P(A)
b) ≤ P(A) d) P(A) minus P(B).

12) Which is always true?

a) P(A) × P(¬A) = 1 c) P(A) + P(¬A) ≤ 1
b) P(A) + P(¬A) = 1 d) P(A) + P(¬A) = 0

13) How is P(A-AND-B) calculated?

a) Multiply probabilities. c) Find the highest probability.
b) Average probabilities. d) Add probabilities.

14) What is P(A-AND-B)?

a) ≥ P(A) c) Between P(A) and P(B).
b) Not enough information to know. d) ≤ P(A)

15) What is the joint probability of two mutually exclusive events?

a) 100% c) 50%
b) 0% d) Not enough information to know.

16) How is the chance of a disjunction of conjunctions calculated?

a) Sum of products. c) Product of sums.
b) Sum of sums. d) Product of products.

17) How many is “not at least 1”?

a) All c) None
b) > 1 d) 1

- 139 -
 -PRACTICE EXAMS-

PRACTICE EXAM 1

1) Answer the questions based on the story. Show your calculation.

The FRAZ-3 virus is spreading across Azmakia and the country is racing to develop a vaccine
that is both safe and effective. Azmakia's labs have been developing several vaccines and now
5 are ready for human trials. Based on past trials, the Health Department knows that a trial
vaccine has a 3/4 chance of being safe and a 1/3 chance of being effective.

a) What's the chance that the first vaccine is safe or effective?

b) What's the chance that any of the vaccines are effective?
c) What’s the chance that all the vaccines are safe?
d) They test 4 vaccines and none are effective. The Health Department says: "Well, after that,
this last vaccine is sure to be effective." Evaluate the Health Department's reasoning.

2) Evaluate the reasoning.

Azmakia has been battling an epidemic of the new FRAZ-3 virus. The country finally has a
vaccine and the government has been making good progress getting the population vaccinated.
Most people have been vaccinated. But now the government gets some shocking data from
the hospitals: 80% of infected people are vaccinated (an infected person is 80% likely to have
been vaccinated)! The government is upset – they think that the vaccine doesn’t work well.

3) Answer the questions based on the story. Show your calculation.

Abby and Bob have kept ratings of their meals at two restaurants: The Lemon and Antique
Land. Today they want the best meal. Which restaurant should they choose?

At The Lemon, 10% of their meals are 1 star, 20% are 2 stars, and 70% are 3 stars, and none
are 4 stars. At Antique Land, 20% of their meals are 1 star, 30% are 2 stars, 10% are 3 stars,
30% are 4 stars, and the other times it’s closed without notice (0 stars).

- 140 -
 PRACTICE EXAM 2

1) Answer the questions based on the story. Show your calculation.

Two viruses begin to spread in Azmakia: FRAZ-3 and GLOP-2. The Health Department wants
to prioritize a vaccine for the virus they expect to infect more people. What should they do?

With any virus, an infected person could be a Superspreader. With FRAZ-3, the chance of being
a Superspreader is 5%; with GLOP-2, the chance is 10%. FRAZ-3 Superspreaders infect an
average of 25.5 other people, while Normals infect 1.6 others. GLOP-2 Superspreaders infect
10.9 others, while Normals infect 2.1 others.

2) Answer the questions based on the story. Show your calculation.

Wackadoodle Industries has a great new product design and they’re concerned that a member
of the development team will betray the company by selling the design to Wackadoodle’s
competitor. They want to assign the product’s development to the team with the lowest chance
of betrayal. Which team should the designers choose?

Team A is loyal but large: each member has a 0.25% chance of betrayal, but there are 60
members. Team B is small but less loyal: 22 members, each with a 0.75% chance of betrayal.

3) Evaluate the reasoning.

There’s a large tropical storm in the West Pacific. Probably it’ll become a typhoon. Although
historically only about 1-in-3 of the storms like this one have developed into a typhoon, it’s
been several years since there was a typhoon in the West Pacific, and there’s normally a
typhoon there every year.

4) Answer the questions based on the story. Show your calculation.

Bob and Abby are going to the movies. Bob will choose the movie. What’s the chance that Abby
falls asleep during the movie?

There’s a 30% chance Bob will choose a comedy, a 50% chance he’ll choose an action film,
and a 20% chance he’ll choose a drama. There’s a 20% chance Abby will fall asleep during a
comedy, a 40% chance she’ll fall asleep during an action film, and a 10% chance she’ll fall
asleep during a drama.

- 141 -
 PRACTICE EXAM 3

1) Write the conditional probabilities that correspond to the statistical generalizations. Draw a
diagram with two labelled circles that shows (imagines) both statements true.

Most things that make the world better (TTMWB) are charities (C).
P( | ) = high
Few charities make the world a better place.
P( | ) = low

2) Answer the questions based on the story. Show your calculations.

Abby is a public health inspector for the Gotham Health Department. This week she will
randomly pick 8 restaurants from the database and do surprise inspections. Normally when
the GHD does surprise inspections, they find food safety code violations in 15% of restaurants.

a) What’s the chance that both of the first 2 restaurants Abby inspects have a code violation?
b) What’s the chance that Abby finds any code violations in her inspections this week?

Restaurants in Gotham serve cuisine from several different countries. The most popular food
is Azmakian: 26% of Gotham restaurants. Azmakian restaurants are typical for food safety: no
more and no less likely to have a code violation.

c) What’s the chance that the first restaurant Abby inspects is Azmakian or has a code violation?

Other kinds of restaurants are not typical for food safety. Kazamni restaurants (less popular:
only 9% of Gotham restaurants) are worse. Abby knows from her past inspection data that a
restaurant with a code violation has a shocking 2/5 chance of being Kazamni.

d) What’s the chance that the first restaurant Abby inspects is Kazamni and has a code violation?

e) Abby visits 7 restaurants and all of them have code violations. She thinks: “This is crazy! Well
after that I can be pretty sure my 8th restaurant won’t have a code violation.” Evaluate Abby’s
reasoning.

- 142 -
 -ANSWER KEY-

QUICK TEST
1) b 2) d 3) c 4) d 5) b 6) a 7) d 8) c 9) a 10) a 11) c 12) b 13) a 14) d
15) b 16) a 17) c

PRACTICE EXAM 1

1)
a) Answer
P(Safe-OR-Effective) = P(Safe) + P(Effective) – P(Safe-AND-Effective)
3
= /4 + 1/3 – (3/4 × 1/3)
9
= /12 + 4/12 – 3/12
10 5
= /12 = /6 [Also correct: 0.833]

Explanation (not required for exam answers)

Disjunction rule for non-mutually exclusive events. Subtraction of the joint probability corrects
for overcounting with addition. Since we have no information that effectiveness and safety are
non-independent (nothing says that either event changes the chance of the other), we must
treat them as independent and multiply their probabilities for the joint probability.

b) Answer
P(AL1-Effective(5Vaccines)) = 1 – P(¬Effective)5 = 1 – (1 – P(Effective))5
= 1 – (1 – 1/3)5
32
= 1– /243
211
= /243 [Also correct: 0.868]

Explanation
Any makes this an At Least 1 probability. The chance of any effective equals 1 minus the chance
of none effective. “None effective” is the conjunction of 5 ineffective vaccines. Since every
vaccine has the same chance of being ineffective (they are independent for effectiveness), we
can use an exponent for the multiplication.

- 143 -
 c) Answer
P(Safe(5Vaccines)) = P(Safe)5
= (3/4)5
243
= /1,024 [Also correct: 0.237]

Explanation
Conjunction rule. Vaccines are independent for safety: every vaccine in a human trial has the
same chance of being safe. So we can use an exponent for the multiplication.
d) Answer
This reasoning is the gambler’s fallacy. The vaccines are independent. The chance of
effectiveness is 1/3 for every vaccine. It does not go up after several ineffective vaccines.

2) Answer
Basic answer: The government is confused about conditional probabilities. Most infected people
were vaccinated (P(Vaccinated|Infected) is high). This does not mean that most vaccinated
people get infected (P(Infected|Vaccinated) is high). Maybe (hopefully) very few vaccinated
people get infected.
Extended answer: As more and more people get vaccinated, the percentage of infected people
who were vaccinated must rise since there are fewer and fewer unvaccinated people left. What’s
important (for public health) is that the number of infected people goes down even if the
percentage of them who were vaccinated goes up.

3) Answer
Expected meal rating = (0 Stars × P(0Stars)) + (1 Star × P(1Star)) + (2 Stars × P(2Stars))

+ (3 Stars × P(3Stars)) + (4 Stars × P(4Stars))

The Lemon: (0 × 0) + (1 × 0.1) + (2 × 0.2) + (3 × 0.7) + (4 × 0) = 2.6 stars
Antique Land: (0 × 0.1) + (1 × 0.2) + (2 × 0.3) + (3 × 0.1) + (4 × 0.3) = 2.3 stars
They should choose The Lemon for lunch.

Explanation
Expected values. The units of value are rating stars. For each restaurant, the expected rating is a
sum of products, the frequency-weighted average of the possible ratings. The weights add to 1.
The solution here is fully detailed for clarity. On the exam, you may omit 0-weighted terms.

- 144 -
 PRACTICE EXAM 2

1) Answer
Expected new infections from each infected person = (Normal Infections × P(Normal)) +
(Superspreader Infections × P(Superspreader))
FRAZ-3: (1.6 × 0.95) + (25.5 × 0.05) = 1.52 + 1.275 = 2.795 infections
GLOP-2: (2.1 × 0.9) + (10.9 × 0.1) = 1.89 + 1.09 = 2.98 infections
The Health Department should prioritize vaccine development for GLOP-2.

Explanation
Expected values. The units of value are new infections. For each virus, the solution is the sum of
products, a probability-weighted average of the possible new infections. The weights must add up
to 1. Since each infected person is a Normal or a Superspreader, P(Normal) = 1 –
P(Superspreader).

2) Answer
P(AL1-ABetray) = 1 – (1 – P(ABetray))60
= 1 – (1 – 0.0025)60 ≈ 0.139
P(AL1-BBetray) = 1 – (1 – P(BBetray))22
= 1 – (1 – 0.0075)22 ≈ 0.153
Wackadoodle should assign the product development to team A.

Explanation
Wackadoodle’s competitor gets their new product design if any team member betrays the
company. That’s the probability that Wackadoodle must consider. So this is an At Least 1 question.
Since they want the lowest chance of betrayal, team A is the better choice.

3) Answer
This looks like the gambler’s fallacy. If there is no reason to believe that a recent lack of typhoons
raises the chance of a typhoon this year, the way this storm develops is independent of the recent
weather. The statistics give every storm the same 1-in-3 chance of becoming a typhoon.

4) Answer
P(FallAsleep) = (P(FA|Com) × P(Com)) + (P(FA|Act) × P(Act)) + (P(FA|Dra) × P(Dra))
= (0.2 × 0.3) + (0.4 × 0.5) + (0.1 × 0.2) = 0.28

- 145 -
Explanation
Conjunction and disjunction rules combined. The solution is the sum of products, a probability
weighted average of the probabilities of the possible movie choices. The weights add up to 1, as
they must. The answer, the average of the Fall Asleep probabilities, is between the high (40%)
and low (10%) probabilities.

- 146 -
 PRACTICE EXAM 3

1) Answer

Most things that make the world better (TTMWB) are charities (C).
P(C|TTMWB) = high
Few charities make the world a better place.
P(TTMWB|C) = low

Charities
TTMWB

2)
a) Answer
P(CV(2Restaurants)) = P(CV)2
= (0.15)2
= 0.0225

Explanation (not required for exam answers)

Conjunction rule. Restaurants are independent for code violations: every restaurant has the
same chance of a violation. So we multiply 15% × 15% or use an exponent of 2.

b) Answer
P(AL1-CV(8Restaurants)) = 1 – (1 – P(CV))8
= 1 – (1 – 0.15)8 ≈ 0.728

Explanation
Any makes this an At Least 1 probability. The chance of any violations equals 1 minus the
chance of none. “None” is the conjunction of 8 violations. Since every restaurant has the chance
of a violation (they are independent for food safety), we can use an exponent for the
multiplication.

- 147 -
c) Answer
P(CV-OR-Az) = P(CV) + P(Az) – P(CV-AND-Az)
= 0.15 + 0.26 – (0.15 × 0.26)
= 0.41 – 0.039
= 0.371

Explanation
Disjunction rule for independent events. Subtraction of the joint probability corrects for
overcounting with addition. Since CV is independent of Azmakian, we multiply their
probabilities for the joint probability.

d) Answer
P(Kaz-AND-CV) = P(Kaz|CV) × P(CV)
2
= /5 × 0.15
= 0.06

Explanation
Conjunction rule. Kazamni and code violation are not independent – a violation raises the
chance that a restaurant is Kazamni. We need the conditional probability of Kazamni given
violation. There are two solutions for joint probability. The other solution is P(CV|Kaz) ×
P(Kaz). However, although we have P(Kaz), we don’t know P(CV|Kaz) (the chance that a
Kazamni restaurant has a violation), P(CV|Kaz), so this solution is not usable.

e) Answer
This reasoning is the gambler’s fallacy. The restaurants are independent. The chance of a
violation is 15% for every restaurant. It does not go up after several restaurants with violations.

- 148 -
 ~7~
-BAYES’ RULE-

i. -BAYES’ RULE-

Two things matter when we judge the probability of a hypothesis based on some new evidence:

1) Prior Probability of the Hypothesis

When new evidence E gives H a new probability, what is that new

probability? This depends on how likely H was before the new
evidence, its prior probability. This is the plausibility P(H)
(believability) of H, based on background information, or the base
rate or prevalence (how often H is true in this kind of situation).

2) Strength of the Evidence

Evidence strength is the power of the evidence to change the probability of the hypothesis. Evidence
strength is the ratio of two probabilities:

Chance of Evidence E if Hypothesis H is true P(E|H)

Chance of Evidence E if Hypothesis H is false P(E|¬H)

Evidence strength > 1: E confirms. E raises the chance of H.

Evidence strength < 1: E disconfirms. E lowers the chance of H.

We need a tool to update the probability of a hypothesis from its prior probability based on the strength
of some new evidence. This tool is Bayes’ rule. Although the rule is mathematical, it teaches some
general lessons that are valuable even without a calculation. We’ve just learned the first two lessons:

BAYES LESSON #1: Consider the prior probability.

BAYES LESSON #2: Confirming E: more likely if H is true, less likely if H is false.

- 149 -
 Bayes’ Rule

Bayes’ rule for updating the chance of H based on E combines the AND and OR rules we learned in
Unit 6. The rule has a short form and an expanded form.

The conjunction rule for Hypothesis-AND- P(H|E) × P(E) = P(E|H) × P(H)

Evidence can be written two ways:

To get P(H|E), we divide both sides by P(E). P(H|E) × P(E) = P(E|H) × P(H)
On the left side, P(E) cancels. P(E) P(E)

This is the short form of Bayes’ rule. P(H|E) = P(E|H) × P(H)

P(E)

If E happens, it happens in one of two situations: H true or H false. To show the overall chance of E,
we can expand the P(E) denominator using the AND (multiplication) and OR (addition) rules.

E: (E-AND-H) - OR - (E-AND-(¬H))

P(E) = (P(E|H) × P(H)) + (P(E|¬H) × P(¬H))

The expanded form of Bayes’ rule uses this expanded denominator.

Bayes’ Rule

(P(E|H) × P(H))
P(H|E) =
(P(E|H) × P(H)) + (P(E|¬H) × P(¬H))

In the denominator, P(E|H) and P(H) are copied from the numerator; and P(¬H) equals 1 – P(H), so
it’s also based on the numerator. As with the short form of Bayes’ rule, we need three probabilities:
P(E|H) and P(E|¬H) for the evidence strength, and the prior probability P(H).

- 150 -
 How to Draw a Bayes Box

A Bayes box is a diagram constructed with the numbers used in Bayes’ rule. It’s a simple way to
show an updated probability as an odds ratio.

Each side of the box equals 1 (100%).

We’ll put 0 on the left, at the bottom.

0 1

Here is an example Bayes box.

We’ll fill in the box in three steps.

STEP 1
What are the prior odds of H?
(Divide the box between P(H) and P(¬H).)
(H) (¬H)

STEP 2 (¬E|H)
What’s the chance of E if H is true?

(E|H)

STEP 3 (¬E|¬H)
What’s the chance of E if H is false?

(E|¬H)

- 151 -
 Example Bayes Box

(¬E|H)
(¬E|¬H)

(E|H)
(E|¬H)

(H) (¬H)

How to Look at a Bayes Box

The example Bayes box above is drawn to show prior odds of H of 1-to-3.

1 - to - 3

So the prior probability of H is 1-in-(1+3) = 1/4.

When the box is filled in with the Evidence probabilities P(E|H) and P(E|¬H), we get two pairs of
rectangles: a shaded pair for E, and an empty (unshaded) pair for ¬E. We’ll first compare the E
rectangles to update H on E; then we’ll compare the ¬E rectangles to update H on ¬E.

- 152 -
1) Hypothesis given Evidence

The Bayes box shows the updated odds of Hypothesis given Evidence. We compare the size (area) of
the H rectangle to the size of the ¬H rectangle.

3
Updated odds of H = - to -

After updating H on E, our visual estimate gives approximately 2:3 odds.

Bayes’ rule converts these updated odds into a chance: divide the H rectangle by the total (sum).

P(H|E) ≈ = 2
/5

+ 3

This is confirming evidence because P(E|H) > P(E|¬H). E raises the chance of H from ~1/4 (prior) to
~2/5 (updated).

A quickly sketched Bayes box is a simple way to double-check a calculated answer. It should also give
you a feel for how Bayes’ rule works. Each probability in the rule is one dimension of a rectangle. The
H rectangle size compared to the ¬H rectangle size is the updated odds.

- 153 -
2) Hypothesis given ¬Evidence

The same Bayes’ box also shows the updated odds of H given ¬E. We compare the two empty
(unshaded) rectangles.

Updated odds of H = 1 - to -
6

A visual estimate gives us approximately 1:6 odds.

Bayes’ rule converts the updated odds into a chance: divide the H rectangle by the total (sum).

1
P(H|¬E) ≈ = 1
/7

1 +
6

This is disconfirming evidence because P(¬E|H) < P(¬E|¬H). It lowers the chance of H from ~1/4
(prior) to ~1/7 (updated).

With these two updates, we see:

BAYES LESSON #3: If E confirms H then ¬E disconfirms H.

In Unit 5 we learned about confirmation bias. This bias is a failure to understand Bayes Lesson #3 –
in particular, when someone gives an ad hoc explanation of disconfirming evidence. They accept E as
confirmation but don’t accept ¬E as disconfirmation.

- 154 -
 Evidence Strength

It’s convenient to have a word as well as a number for evidence strength. Terms like these will be
suitable for our examples:

Strong Moderate Weak Neutral Weak Moderate Strong

0 1
/10 1
/5 1
/2 1 2 5 10 ∞

If P(E|H) (numerator) is 0, P(E|H) As P(E|¬H) (denominator)

the value of the ratio is 0: P(E|¬H) goes to 0, the value of the
maximum disconfirmation. ratio goes to infinity:
maximum confirmation.

We learned (Bayes Lesson #3) that if E confirms then ¬E disconfirms. However this does not mean
that the confirming strength of E must be equal to the disconfirming strength of ¬E. For example, in
the case below, E confirms more strongly than ¬E disconfirms.

0.4 1
Disconfirming strength of ¬E: /0.8 = /2

0.4
0.8
0.6

0.2

0.6
Confirming strength of E: /0.2 = 3

- 155 -
 Tests

A test asks: Is this hypothesis true?

A Yes answer is a positive result; an incorrect Yes is a false positive (“false alarm”).
A No answer is a negative result; an incorrect No is a false negative (“miss”).

Sensitivity is the true positive rate, how well the test detects when the hypothesis is true.
Sensitivity = 1 – false negative rate

Specificity is the true negative rate, how well the test detects when the hypothesis is false.
Specificity = 1 – false positive rate

False Negative Rate Specificity

(True Negative Rate)

Sensitivity
(True Positive Rate) False Positive Rate

Base Rate

A test that is both sensitive and specific is also accurate – most of its results are true.* A perfect test
would have zero errors of either kind, so its results (positive or negative) would always be true.

*
The accuracy – the percentage of all results that are true – is somewhat dependent on the base rate, but any test that has
high sensitivity and high specificity has at least fairly high accuracy. In contrast, the positive predictive value (PPV) – the
chance that H is true given a positive result (the updated probability of H) – is very dependent on the base rate (the prior
probability of H). A test can have high accuracy and low PPV if, due to a low base rate, most results are true negatives but
most positive results are false positives.

- 156 -
ii. -BAYESIAN UPDATING – VISUAL EXAMPLE-

This section uses the 20 UFOs from Unit 6 for an example where we can see the answers in a picture.

Xenu Zargon Other Planet Abducting Non-abducting

(5) (8) (7) (6) (14)

A UFO lands. The aliens tell us “We’re from Xenu.” Will they
abduct us? What is the updated probability P(Abduct|Xenu)?

STEP 1
The prior probability the
6
aliens plan to abduct us: /20

(Abd) (¬Abd)

STEP 2
The chance the UFO (¬Xen|Abd)
is from Xenu if they
1
plan to abduct us: /6 (Xen|Abd)

STEP 3
The chance the UFO is (¬Xen|¬Abd)
from Xenu if they do
4
not plan to abduct us: /14 (Xen|¬Abd)

- 157 -
The first thing to notice is that the Xenu is disconfirming evidence: P(Xen|Abd) < P(Xen|¬Abd). We
expect P(Abd|Xen) (updated) < P(Abd) (prior).

4
Updated odds of Abduct = 1 - to -

A visual estimate gives us updated odds of approximately 1:4.

Bayes’ rule converts the updated odds into a chance: divide the Abduct rectangle by the total (sum).

P(Abd|Xen) ≈ = 1
/5
+ 4
1

(1/6 × 6
/20)

(P(Xen|Abd) × P(Abd))
P(Abd|Xen) = = 1
/5
(P(Xen|Abd) × P(Abd)) + (P(Xen|¬Abd) × P(¬Abd))

(1/6 × 6
/20) + (4/14 × 14
/20)

The calculated probability matches the visual

estimate. Probably we calculated correctly. We can
also see the correct answer in the original picture:
1-in-5 Xenu UFOs abduct.

The updated probability can help us decide how to act. Suppose we think it would be very cool to meet
space aliens, but we’re afraid of being abducted. We think we should stay and meet the aliens if and
only if the chance of abduction is less than 25%. If it’s at least 25%, we should run away. Based on
the Xenu evidence, the chance is now less than 25%. We should meet the aliens!

- 158 -
Suppose instead that the aliens come out of the UFO and say “We’re not from Xenu.” This is confirming
evidence: P(¬Xen|Abd) > P(¬Xen|¬Abd). We expect P(Abd|¬Xen) > P(Abd).

2
Updated odds of Abduct = 1 - to -

A visual estimate gives us updated odds of approximately 1:2.

Bayes’ rule converts the updated odds into a chance: divide the Abduct rectangle by the total (sum).

P(Abd|¬Xen) ≈ = 1
/3

+
2
1

(5/6 × 6
/20)

(P(¬Xen|Abd) × P(Abd))
P(Abd|¬Xen) = = 1
/3
(P(¬Xen|Abd) × P(Abd)) + (P(¬Xen|¬Abd) × P(¬Abd))

(5/6 × 6
/20) + (10/14 × 14
/20)

The calculated answer matches

the visual estimate and the
picture: 5-in-15 (1/3) non-Xenu
UFOs abduct. Since this updated
probability of abduction is at
least 25%, we should run away!

- 159 -
iii. -KEEP UPDATING!-

A UFO lands and it’s from Zargon. We think the aliens may plan to
abduct us. We know other information that allows us to test our
hypothesis: the space gun test. Do the aliens have space guns?

The test is fairly sensitive: 70% of abducting aliens have space guns to zap humans. Only 30% do not
(the test’s false negative rate). The test is also quite specific: 90% of non-abducting aliens do not
have space guns. Only 10% of non-abducting aliens have space guns (the false positive rate).

These Zargon aliens have space guns! We can update the probability of Abduct a second time. The
original prior probability of Abduct was 6/20. Updated on the Zargon evidence, the chance of Abduct
increased to 3/8. That becomes the new prior probability for updating on the Space Gun evidence.

(P(Z|A) × P(A)) (3/6 × 6/20)

P(Abd|Zar) = = = 3
/8
3 6 5 14
(P(Z|A) × P(A)) + (P(Z|¬A) × P(¬A)) ( /6 × /20) + ( /14 × /20)

(P(SG|A) × P(A|Z)) (0.7 × 3/8)

≈ 0.81
=
P(Abd|SG) =
(P(SG|A) × P(A|Z)) + (P(SG|¬A) × P(¬A|Z)) (0.7 × 3/8) + (0.1 × 5/8)

Here’s our last lesson of Bayesian hypothesis testing:

BAYES LESSON #4: Yesterday’s updated probability = today’s prior probability.

- 160 -
iv. -SUBJECTIVE PROBABILITY-

This introduction to Bayes’ rule uses probabilities that are directly calculated from a simple picture to
make them as clear as possible. But many probabilities cannot be assigned in this way.

What’s the chance?

• There is advanced intelligent life elsewhere in
our galaxy.
• Qatar bribed FIFA to be made host of the 2022
World Cup.
• Canada’s next prime minister is a woman given (Image: medium.com)

that they belong to the Conservative party.

There is no way to calculate probabilities such as these. We cannot count equally likely possible
outcomes and there don’t seem to be any frequencies that directly give us the answers. (For example,
there was only one 2022 World Cup.) But we can give a subjective probability that expresses a
degree of belief (confidence or feeling of certainty).

Even if they aren’t mathematically correct or incorrect, subjective probabilities can still be reasonable
or unreasonable. We can explain or justify this kind of probability, but normally we’ll be explaining
why it is, for example, 60% rather than 80% or 40%, not 60% rather than 61% or 59%.

Even though subjective probabilities do not come directly from counting or measuring anything, they
are real probabilities! The rules of probability covered in Units 6-7 apply to them just as they do to
other probabilities. The rules tell us what degree of belief we should have in something given the
degrees of belief we have in other things.

What’s the chance?

How to assign a probability to H? First think of some commonly used subjective probabilities in
different ways, as in the chart below.

- 161 -
 Percentage Fraction Odds
100% 1 1:0 For the smallest benefit, you would bet anything that
hypothesis H is true. This probability says that no evidence
at all would ever change your mind about the truth of H.
99
99% /100 99:1 For many things, we act as though “99% sure” means “no
doubt”, although for very important things (e.g. my plane
will not crash) we often want even higher confidence (e.g.
99.9%, 99.999%...).
19
95% /20 19:1 This probability is high enough to be considered significant
for some common statistical purposes (see Unit 9).
4
80% /5 4:1
3
75% /4 3:1
2
66.6% /3 2:1
1
50% /2 1:1 If you had to guess whether H is true or false, you would
just flip a coin to choose your answer.
1
33.3% /3 1:2
1
25% /4 1:3
1
20% /5 1:4
1
5% /20 1:19
1
1% /100 1:99
0% 0 0:1

Here are two ways to discover your degree of belief in H.

Fair Bets
You can measure your degree of belief by asking yourself what a fair bet would be – neither side of
the bet has an advantage, so you wouldn’t prefer to take one side of the bet more than the other. For
example, maybe you feel that this bet is fair: win $1 if H is true; lose $2 if H is false. This means you
think the odds of H are 2-to-1 (since $1 × 2 = $2 × 1). Or maybe it seems fair to win $5 if H is true
and lose $1 if it’s false. Then you think the odds of H are 1-to-5 (since $5 × 1 = $1 × 5).

Imaginary Frequencies
Imagine that the evidence relevant to H exists in many imaginary “worlds”, or that many imaginary
people have the same evidence that you have. How often in all of these worlds, or for all these people,
do you think the hypothesis is true? This frequency tells you your subjective probability.

- 162 -
 -UNIT 7 SKILLS-

You must be able to:

• Write Bayes’ rule for a hypothesis and evidence.

• Recognize and use base rates and test error rates.

• Draw a Bayes box and visually estimate the updated probability of H based on E and based on ¬E.

• Calculate a Bayesian update using Bayes’ rule.

- 163 -
 -QUICK TEST QUESTIONS-

1) What is evidence strength?

a) P(E|H) / P(¬E|¬H) c) P(E|H) / P(¬E|H)
b) P(E|H) / P(E|¬H) d) P(E|¬H) / P(E|H)

2) To update the probability of a hypothesis based on evidence, what do we need to know?

a) Simplicity of the evidence. c) Prior probability of the evidence;
b) Prior probability of the hypothesis; strength of the hypothesis.
strength of the evidence. d) Plausibility of the evidence.

3) How many independent probabilities (probabilities that we need to separately learn) are in the
expanded form of Bayes’ rule?
a) 6 c) 4
b) 5 d) 3

4) When does evidence E confirm hypothesis H?

a) P(E|H) > P(E|¬H) c) P(H) > P(¬H)
b) P(H) > P(E) d) P(E|¬H) > P(E|H)

5) When is hypothesis H confirmed by evidence E?

a) P(H) > P(E) c) P(H|E) > 0.5
b) P(H|E) > P(H) d) P(E|¬H) > P(E|H)

6) Label the Bayes box.

A D
B C
a) A: (¬E|H); B: (H); c) A: (E|H); B: (H);
C: (¬H); D: (¬E|¬H) C: (¬H); D: (E|¬H)
b) A: (H); B: (E|H); d) A: (E|H); B: (H|E);
C: (E|¬H); D: (¬H) C: (E|¬H); D: (¬H|E)

- 164 -
7) Choose the correct statement about a test.
a) Sensitivity, specificity, false negative rate, c) Sensitivity = 1 – specificity; false
and false positive rate are all independent. negative rate = 1 – false positive rate
b) Sensitivity = 1 – false positive rate; d) Sensitivity = 1 – false negative rate;
specificity = 1 – false negative rate specificity = 1 – false positive rate

8) Which is determined by a base rate?

a) Evidence strength c) True positive rate
b) Prior probability d) Test specificity

9) Which parts of the Bayes box are compared for updating on ¬E (false prediction/negative result)?
P R

Q S

a) P and R c) P and Q
b) Q and S d) R and S

10) Which probability is determined by the false negative rate of a test?

a) P(E|¬H) c) P(H|¬E)
b) P(¬E|H) d) P(¬E|¬H)

11) When is checking for evidence E an accurate test of hypothesis H?

a) P(E|H) = low; P(E|¬H) = low c) P(E|H) = low; P(E|¬H) = high
b) P(E|H) = high; P(E|¬H) = high d) P(E|H) = high; P(E|¬H) = low

12) Which part of a Bayes box comes from the false positive rate of a test?

A C
B D

a) A c) C
b) B d) D

13) To update on an additional piece of evidence, the previous ______ becomes the new _______.
a) prior probability… updated probability c) sensitivity… specificity
b) updated probability… prior probability d) false positive rate… false negative rate

- 165 -
 -PRACTICE EXAMS-

PRACTICE EXAM 1

Answer the questions based on the story.

Crustie’s auction guarantees the authenticity of items they

auction. Their policy is to auction an item only if it has at least
15-to-1 odds of being authentic, as determined by their lab.

A seller brings a violin that he claims was made in 1680 by Leonardo Varistradi. In the past,
Crustie’s has found that 60% of claimed Varistradi violins are authentic. Another ¼ are made by
apprentices in Varistradi’s workshop (less valuable). The rest are forgeries (worthless fakes).

With a microscope, the lab can see what kind of wood a violin is
made of. Varistradi usually made violins with Mieffe Valley Cyprus.
This can be used as a test for authenticity. The MVC test is 75%
sensitive. And the MVC test has false positives: 10% of the
inauthentic violins are MVC by workshop apprentices, and an
additional 8% of inauthentic violins are MVC by forgers.

1) Write the information for the Mieffe Valley Cyprus (MVC) test for an Authentic Varistradi (AV).
AV base rate: False negative rate: Specificity:
Sensitivity: False positive rate:

2) Draw a labelled Bayes box for the AV hypothesis and the MVC evidence.

3) Visually estimate the updated probability of AV after each possible test result.
Probability of AV after positive result (MVC): Odds:______ Chance:_______
Probability of AV after negative result (¬MVC): Odds:______ Chance:_______

The lab checks the violin. It’s not made of MVC wood.

4) Write Bayes’ Rule (just the rule, no numbers) for the updated probability of AV.

5) Calculate the updated probability of AV. Show your calculation.

STOP. Does your calculated answer match your visual estimate?

- 166 -
6) The lab’s analysis confirms | disconfirms the AV hypothesis because…

7) Now Crustie’s should | should not auction the violin because…

8) Imagine instead that the violin wood is MVC. Use Bayes’ Rule to calculate what the updated
probability of AV would be. Show your calculation.
STOP. Does your calculated answer match your visual estimate?

The lab looks inside the violin at the maker’s label. It’s an unusual bumpy paper that Varistradi
liked. This is important because only 10% of authentic Varistradi violins are missing this bumpy
label, but the vast majority – 98% – of inauthentic violins do not have it.

9) Based on this new, additional evidence, what is the new updated probability of the AV
hypothesis? Show your calculation.

10) Now Crustie’s should | should not auction the violin.

- 167 -
 PRACTICE EXAM 2

Answer the questions based on the story.

Archaeologists are studying the early settlement of the Azmakia region. They think there’s a ¼
chance that the first settlers were the Gthylio people. But most likely – they think 70% – the
Trameron people arrived first by boat. There are also some other unlikely theories.

There’s an ancient painting in the Wazoo cave.

Archaeologists know that only about 20% of
Trameron paintings are made with ochre (natural
paint), but ochre is more common in other
ancient cave paintings: about 55% of those.

They send a sample of the cave paint to a lab. They have a small research budget; they decide
to use it to search the coast for Trameron artifacts if and only if the odds of the Trameron
hypothesis become at least 3-to-2.

1) Draw a labelled Bayes box for the Trameron (Tram) hypothesis and the Ochre (Och) evidence.

2) Visually estimate the updated probability of Tram after each possible lab result.
Probability of Tram given Och: Odds:______ Chance:_______
Probability of Tram given ¬Och: Odds:______ Chance:_______

The lab announces the findings of the analysis. The Wazoo paint sample contains ochre.

3) Write Bayes’ Rule (just the rule, no numbers) for the updated probability of Tram.

4) Calculate the updated probability of Tram. Show your calculation.

STOP. Does your calculated answer match your visual estimate?

5) The lab’s analysis confirms | disconfirms the Trameron hypothesis because…

6) Now the archaeologists should | should not use their budget to search the coast because…

7) Imagine instead that the paint had not contained ochre. Use Bayes’ Rule to calculate what the
updated probability of Tram would be. Show your calculation.
STOP. Does your calculated answer match your visual estimate?

- 168 -
 Pre-historic painters sometimes made hand outlines. There’s a
hand outline test for a Trameron painter. It’s not very sensitive
since 2-in-8 Trameron paintings never had hand outlines and
experts think that another 1-in-8 did have one, but it faded away.
However the test is quite specific since the Trameron were almost
the only painters who made hand outlines. In 95% of paintings by
other pre-historic peoples, the painters never made hand outlines.

8) Write the information for the Hand Outline (HO) test for Tram.
False negative rate: Specificity:
Sensitivity: False positive rate:

Looking carefully around the cave painting, the archaeologists are excited to find a hand outline.

9) Based on this new, additional evidence, what is the new updated probability of the Tram
hypothesis? Show your calculation.

10) Now the archaeologists should | should not use their budget to search the coast.

- 169 -
 PRACTICE EXAM 3

Answer the questions based on the story.

Ozville has a tornado warning siren. There’s no tornado tonight

but the siren sounds. City officials’ best explanation is that a
electrical malfunction activated the siren. An alternative is that
computer hacker criminals sounded the siren to scare people.

The system has never malfunctioned, and hackers have hacked other emergency systems. Based
on this, the officials think that a malfunction is 60% likely and a hack 30% likely.

The city can upgrade the sirens with expensive encryption to prevent hacking. They want to
spend the money for that if and only if the odds of a hack are at least 1-to-2.

Hackers would have only a 1/4 chance of breaking the access

panel without scratching it. But raccoons sometimes scratch
the panel. It’s been a while since officials last checked and by
now there’s about a 35% chance the panel has been scratched
by raccoons even if the sirens weren’t hacked.

1) Draw a Bayes box for the Hack hypothesis and the Scratched Access Panel (SAP) evidence.

2) Visually estimate the updated probability of Hack (odds and chance) after each possible
discovery at the access panel.
Probability of Hack given SAP: Odds:______ Chance:_______
Probability of Hack given ¬SAP: Odds:______ Chance:_______

The officials inspect the siren and find that its access panel is scratched.

3) Write Bayes’ Rule (just the rule, no numbers) for the updated probability of Hack.

4) Calculate the updated probability of Hack. Show your calculation.

STOP. Does your calculated answer match your visual estimate?

5) The scratched access panel confirms | disconfirms the Hack hypothesis. What is the evidence
strength?

- 170 -
6) The city should | should not upgrade the siren system with encryption because…

7) Imagine instead that the access panel had not been scratched. Use Bayes Rule to calculate
what the updated probability of Hack would be.
STOP. Does your calculated answer match your visual estimate?

Hackers like to brag in internet forums. A forum post test for a hack is only about 60% sensitive,
and its false positive rate is fairly high: in about 1-in-3 cases when emergency systems are
activated by a malfunction or other non-criminal reason, someone in a forum falsely claims to
have hacked the system just to brag.

8) Write the information for the Forum Post (FP) test for Hack.
False negative rate: Specificity:
Sensitivity: False positive rate:

Searching hacker forums, the officials find a new post by Hackma$ter69 claiming to have
hacked the Ozville tornado siren.

9) Based on this new, additional evidence (Forum Post), what is the new updated probability of
the Hack hypothesis? Show your calculation.

10) The chance of the Malfunction hypothesis does | does not have to change now. What is the
maximum probability of Malfunction now?

- 171 -
 PRACTICE EXAM 4

Answer the questions based on the story.

Zelda wants to be a supermodel. She signs with a modelling

agency that tries to get jobs for her. About 8% of the agency’s
models become supermodels. Another 60% get jobs as regular
models. The rest of the agency’s models give up and leave the
industry after a few months.

The agency has discovered a test for predicting which models will become supermodels. Before they
became supermodels, two thirds (2/3) of supermodels were picked by a local store, FancyFashions,
to be in their advertisements. But some of the non-supermodels who become regular models or leave
the industry after a few months also get picked by FancyFashions: 15% get picked by FancyFashions
just because they’re cheap and available, and another 5% get picked because their parents work for
FancyFashions.

Zelda waits to see if FancyFashions picks her. She has decided that she'll leave school and devote
herself full time to modelling if and only if she has at least 3-to-2 odds of becoming a supermodel.

1) Zelda and her agency want to know if she will become a Supermodel. They use the FancyFashions
Pick test. Give the following information for this test.
Supermodel base rate:
Test false negative rate: Test specificity:
Test sensitivity: Test false positive rate:

2) Fill in a labelled Bayes box for the hypothesis “Supermodel” and the evidence “FancyFashions
Pick”.

3) Visually estimate the updated probabilities for a positive and negative test results.

FancyFashions announces their model picks. They have not picked Zelda.

4) Write Bayes’ rule for the updated probability that Zelda will become a Supermodel, based on
this evidence. (Just the rule. The calculation is in question 5.)

- 172 -
5) The evidence confirms | disconfirms that Zelda will become a supermodel. Explain your
answer. How strong is this evidence? (Calculate the evidence strength.)

6) Calculate the updated probability that Zelda will become a supermodel.

STOP. Does your calculated answer match your visual estimate?

7) Zelda should | should not leave school to model full time. Explain your answer.

8) Imagine instead that FancyFashions does pick Zelda. What would be the updated probability of
her becoming a Supermodel if that happened?
STOP. Does your calculated answer match your visual estimate?

After learning that she didn’t get picked by FancyFashions, Zelda gets an excited message from
her agency. The fashion brand Coolio Filbastra wants her in their next fashion show! The agency
estimates that a non-supermodel would have only a 1-in-200 chance of being chosen by CF. And
CF is a major brand – a third of all supermodels modelled for CF.

9) What’s the updated probability of Zelda becoming a supermodel now?

10) Zelda should | should not leave school to model full time.

11) Briefly explain how the four “Bayes Lessons” of Unit 7 apply to this practice exam, referring to
specific questions.

- 173 -
 -ANSWER KEY-

QUICK TEST
1) b 2) b 3) d 4) a 5) b 6) c 7) d 8) b 9) a 10) b 11) d 12) d 13) b

PRACTICE EXAM 1

1) Answer
AV base rate: 60% False negative rate: 25% Specificity: 82%
Sensitivity: 75% False positive rate: 18%

Explanation
Sensitivity + false negative rate = 1. Specificity + false positive rate = 1.
The question gives us the test sensitivity of 75%. The false negative rate = 1 – sensitivity =
25%. The question gives us the false positive rate in two components (10% and 8%) which we
add together for the total false positive rate of 18%. Specificity = 1 – false positive rate = 82%.

2) Answer

(¬MVC|AV)

(¬MVC|¬AV)
(MVC|AV)

(MVC|¬AV)
(AV) (¬AV)

3) Answer
Visual estimates:
Probability of AV after positive result (P(AV|MVC)): 7:1 = 7/8
Probability of AV after negative result (P(AV|¬MVC)): 1:2 = 1/3

- 174 -
 Explanation
Positive result (MVC): shaded rectangles. The AV shaded rectangle (left) looks way larger than
the non-AV shaded rectangle (right).
Negative result (¬MVC): empty rectangles. The AV empty rectangle (left) looks about half the
size of the non-AV empty rectangle (right).

4) Answer
(P(¬MVC|AV) × P(AV))
P(AV|¬MVC) =
(P(¬MVC|AV) × P(AV)) + (P(¬MVC|¬AV) × P(¬AV))

Explanation
The MVC test is negative – the wood is not MVC. We update AV on ¬MVC.

5) Answer
(0.25 × 0.6)
P(AV|¬MVC) = ≈ 0.314
(0.25 × 0.6) + (0.82 × 0.4)

Explanation
Our visual estimate for AV updated on a negative result (¬MVC) was 1/3 (0.333), so 0.314 is
reasonable.

6) Answer
The lab’s analysis disconfirms the AV hypothesis because P(AV|¬MVC) < P(AV). (The updated
probability is less than the prior.)
[Also correct: …because P(¬MVC|AV) < P(¬MVC|¬AV). (The evidence strength is less than 1.)]

7) Answer
Now Crustie’s should not auction the violin because P(AV|¬MVC) is less than 15-to-1 odds
(15/16).

- 175 -
8) Answer
(0.75 × 0.6)
P(AV|MVC) = ≈ 0.862
(0.75 × 0.6) + (0.18 × 0.4)

Explanation
We’re imagining a positive result with the MVC test. We update AV on MVC. Our visual estimate
for AV updated on a positive result (MVC) was 7/8 (0.875), so 0.862 is reasonable.

9) Answer
(0.9 × 0.314)
P(AV|BumpyLabel) = ≈ 0.954
(0.9 × 0.314) + (0.02 × 0.686)

Explanation
The story continues with a second piece of evidence (the bumpy label). The prior probability of
AV when they check the label is the updated probability of AV after the ¬MVC evidence: 0.314.

10) Answer
Now Crustie’s should auction the violin.

- 176 -
 PRACTICE EXAM 2

1) Answer

(¬Och|¬Tram)

(¬Och|Tram)

(Och|¬Tram)
(Och|Tram)
(Tram) (¬Tram)

2) Answer
Visual estimates:
P(Tram|Och): 5:6 = 5/11
P(Tram|¬Och): 4:1 = 4/5

Explanation
Ochre evidence: shaded rectangles. The Trameron shaded rectangle (left) looks slightly smaller
than the non-Trameron shaded rectangle (right).
¬Ochre evidence: empty rectangles. The Trameron empty rectangle (left) looks way larger than
the non-Trameron empty rectangle (right).

3) Answer
(P(Och|Tram) × P(Tram))
P(Tram|Och) =
(P(Och|Tram) × P(Tram)) + (P(Och|¬Tram) × P(¬Tram))

Explanation
The paint contains Ochre, so we update Tram on Och.

- 177 -
4) Answer
(0.2 × 0.7)
P(Tram|Och) = ≈ 0.459
(0.2 × 0.7) + (0.55 × 0.3)

Explanation
Our visual estimate for Tram updated on Och was 5/11 (0.454), so 0.459 is reasonable.

5) Answer
The lab’s analysis disconfirms the Trameron hypothesis because P(Tram|Och) < P(Tram). (The
updated probability is less than the prior.)
[Also correct: …because P(Och|Tram) < P(Och|¬Tram). (The evidence strength is less than 1.)]

6) Answer
Now the archaeologists should not use their budget to search the coast because the updated
probability of the Trameron hypothesis is less than 3-to-2 odds (3/5).

7) Answer
(0.8 × 0.7)
P(Tram|¬Och) = ≈ 0.806
(0.8 × 0.7) + (0.45 × 0.3)

Explanation
Our visual estimate for Tram updated on ¬Och was 4/5 (0.8), so 0.806 is reasonable.

8) Answer
False negative rate: 3/8 Specificity: 95%
5
Sensitivity: /8 False positive rate: 5%

Explanation
Sensitivity + false negative rate = 1. Specificity + false positive rate = 1.
The story describes two kinds of false negatives (Trameron paintings with no hand outline). We
add these rates together for the total false negative rate. Sensitivity = 1 – false negative rate =
5
/8. The question gives us the test specificity. False positive rate = 1 – specificity = 5%.

- 178 -
9) Answer
(5/8 × 0.459)
P(Tram|HO) = ≈ 0.914
(5/8 × 0.459) + (0.05 × 0.541)

Explanation
The story continues with a second piece of evidence (the hand outline). The prior probability of
Tram when they check for the hand outline is the updated probability of Tram after the Och
evidence: 0.459.

10) Answer
Now the archaeologists should use their budget to search the coast.

- 179 -
 PRACTICE EXAM 3

1) Answer

(¬SAP|Hack)
(¬SAP|¬Hack)

(SAP|Hack)
(SAP|¬Hack)

(Hack) (¬Hack)

2) Answer
Visual estimates:
P(Hack|SAP): 1:1 = 1/2
P(Hack|¬SAP): 1:7 = 1/8

Explanation
SAP evidence: shaded rectangles. The Hack shaded rectangle (left) looks about the same size
as the non-Hack shaded rectangle (right).
¬SAP evidence: empty rectangles. The Hack empty rectangle (left) looks way smaller than then
non-Hack empty rectangle (right).

3) Answer
(P(SAP|Hack) × P(Hack))
P(Hack|SAP) =
(P(SAP|Hack) × P(Hack)) + (P(SAP|¬Hack) × P(¬Hack))

Explanation
The access panel is scratched, so we update Hack on SAP.

- 180 -
4) Answer
(3/4 × 0.3)
P(Hack|SAP) = ≈ 0.479
(3/4 × 0.3) + (0.35 × 0.7)

Explanation
The ¼ chance that hackers would avoid scratching the panel means there’s a ¾ chance they
1
would scratch it. Our visual estimate for Hack updated on SAP was /2 (0.5), so 0.479 is
reasonable.

5) Answer
The scratched access panel confirms the Hack hypothesis because P(SAP|Hack) >
P(SAP|¬Hack). [Also correct: P(SAP|Hack) > P(SAP|¬Hack). (The updated probability is
greater than the prior.)] This is weakly confirming evidence: (3/4 / 0.35) ≈ 2.1.

6) Answer
The city should upgrade the siren system with encryption because the updated probability of
the Hack hypothesis is greater than 1-to-2 odds (1/3).

7) Answer
(1/4 × 0.3)
P(Hack|¬SAP) = ≈ 0.142
(1/4 × 0.3) + (0.65 × 0.7)

Explanation
Our visual estimate for Hack updated on ¬SAP was 1/8 (0.125), so 0.142 is reasonable.

8) Answer
False negative rate: 40% Specificity: 2/3
Sensitivity: 60% False positive rate: 1/3

Explanation
Sensitivity + false negative rate = 1. Specificity + false positive rate = 1.
The information in the question gives us the sensitivity and the false positive rate.

- 181 -
9) Answer
(0.6 × 0.479)
P(Hack|FP) = ≈ 0.62
(0.6 × 0.479) + (1/3 × 0.521)

Explanation
The story continues with a second piece of evidence (Forum Post). The prior probability of Hack
when they check the forums is the updated probability of Hack after the SAP evidence: 0.479.

10) Answer
The chance of the Malfunction hypothesis does change now. Its maximum probability now is
about 38% (1 – 0.62).

- 182 -
 PRACTICE EXAM 4

1) Answer
Supermodel base rate: 0.08
Test false negative rate: 1/3 Test specificity: 0.8
2
Test sensitivity: /3 Test false positive rate: 0.2

Explanation
Sensitivity + false negative rate = 1. Specificity + false positive rate = 1.
The information in the question gives us the sensitivity and the false positive rate. There are two
types of false negative results: “cheap and available” and “parents work for FancyFashions”.
These rates (15% and 5%) must be added together.

2) Answer

(¬FFP|SM)
(¬FFP|¬SM)

(FFP|SM)

(FFP|¬SM)
(SM) (¬SM)

3) Answer
Visual estimates:
P(SM|FFP): 2:5 = 2/7
P(SM|¬FFP): 1:25 = 1/26

Explanation
FFP evidence: shaded rectangles. The SM shaded rectangle (left) looks quite a bit smaller than
the non-SM shaded rectangle (right).
¬FFP evidence: empty rectangles. The SM empty rectangle (left) looks way smaller than the
non-SM empty rectangle (right).

- 183 -
4) Answer
(P(¬FFP|SM) × P(SM))
P(SM|¬FFP) =
(P(¬FFP|SM) × P(SM)) + (P(¬FFP|¬SM) × P(¬SM))

Explanation
FancyFashions does not pick Zelda, so we update the Supermodel hypothesis on the Not-
(FancyFashions Pick) evidence.

5) Answer
The evidence disconfirms that Zelda will become a supermodel because P(¬FFP|SM) <
P(¬FFP|¬SM). This is weakly disconfirming evidence: (1/3 / 0.8) = 0.42.

6) Answer
(1/3 × 0.08)
P(SM|¬FFP) = ≈ 0.035
(1/3 × 0.08) + (0.8 × 0.92)

Explanation
Our visual estimate for Supermodel updated on Not-(FancyFashions Pick) was 1/25 (0.04), so
0.035 is reasonable.

7) Answer
Zelda should not leave school to model full time because P(SM|¬FFP) < 3:2 (3/5).

8) Answer
(2/3 × 0.08)
P(SM|FFP) = ≈ 0.225
(2/3 × 0.08) + (0.2 × 0.92)

Explanation
Our visual estimate for the Supermodel hypothesis updated on the FancyFashions Pick evidence
was 2/7 (0.286), so 0.225 is reasonable.

- 184 -
9) Answer
(1/3 × 0.035)
P(SM|CFP) = ≈ 0.707
(1/3 × 0.035) + (1/200 × 0.965)

Explanation
The story continues with a second piece of evidence (Coolio Filbastra Pick). The prior probability
of Supermodel when Zelda’s agency gets the message is the updated probability of Supermodel
after the Not-(FancyFashions Pick) evidence: 0.035.

10) Answer
Zelda should leave school and model full time.

11) Answer
Consider prior the probability. Question 1 asks for the supermodel base rate. This is where we
think about the chance that Zelda will become a supermodel, based just on her being a model
at her agency, before she gets the new evidence about FancyFashions.
Confirming evidence E: More likely if hypothesis H is true, less likely if H is false (and vice versa
for disconfirming evidence). In question 4, we have the evidence that Zelda wasn’t picked by
FancyFashions. Based on the history of the models they pick becoming supermodels or not, the
fact that they don’t pick her makes it less likely that she’ll become a supermodel.
If evidence E confirms hypothesis H then ¬E disconfirms H. In questions 6 and 8, the two
possible decisions of FancyFashions (pick or not pick) push Zelda’s chance of being a
supermodel in opposite directions, either higher (pick) or lower (not pick) than the prior
probability.
Yesterday’s updated probability = today prior probability. In question 9, Zelda has an additional
piece of evidence (the Coolio Filbastra pick). To update on this new evidence, we used the
previous update (based on FancyFashions not picking her) as the prior probability. The new
evidence tells us the opposite of what the first evidence told her, but by using the previous
updated probability, we base the final probability of Zelda being a supermodel on all the
information we have.

- 185 -
 ~8~
-ANALOGIES-

i. -ARGUING FROM ANALOGY-

Analogies and Analogical Arguments

An analogy says that one thing is similar to another.

Analogy 8.1: The Amazon River is like the Nile River.

In an extended analogical argument, an analogy is the intermediate conclusion. The sub-argument

supports the analogy by noting common (shared) features of the items in the analogy.

Analogical Argument Pattern

Main Argument Sub-argument

• Analogy: A is similar to B. • A and B both have [Common Feature 1].

• B has [Feature]. • Etc.

A has [Feature]. Analogy: A is similar to B.

The argument pattern is inductive. Similar does not mean “the same”! Even if A and B are generally
alike, A might not have the feature in the conclusion.

- 186 -
Here’s an example using the rivers analogy.
The feature is “has several tributary rivers that
flow into it.” The Amazon River is compared to
three other rivers; the analogy is supported by
premises noting three common features.

Argument 8.2: ① The Amazon, Nile, Yangtze, and Mississippi are huge rivers flowing for
thousands of kilometres, ② they periodically flood, and ③ they empty into the sea. And
since ④ the Nile, Yangtze, and Mississippi rivers have several tributary rivers that flow
into them, probably ⑤ the Amazon River has several tributary rivers that flow into it, too.

(Image: worldmapblank.com)

The analogy is implicit: ⑥ “The Amazon River is similar to the Nile, Yangtze, and Mississippi rivers.”
The reconstruction below shows the role of each statement in the argument pattern.

Main Argument Sub-argument

①
④ [B, C, D have Feature.] ②
⑥ [Analogy: A is similar to B, C, D] ③

⑤ [A has Feature.] ⑥

- 187 -
ii. -WEAK ANALOGIES-

Like any argument, an analogical argument needs true premises to be a good argument. To criticize
an analogical argument, someone could disagree with any premise. But very often criticism of an
analogical argument criticizes the analogy.

Even though an analogy is a statement, analogies are usually said to be strong or weak rather than
true or false. An analogy is weak if the compared items are significantly different. When is a difference
significant? There’s no rule for how to judge this significance. All we can do is consider what we know
about the topic and judge if we think the analogy may be misleading.

The Amazon River has flesh-eating piranha fish but the other
rivers don’t. Is this difference significant? Can we think of any
“connection” between piranhas and tributaries that would make
the conclusion less likely? Maybe not. Maybe the piranha
difference isn’t very significant, and the analogy, and argument,
are still fairly strong.
(Image: Nihongonihongo)

- 188 -
iii. -ANALOGICAL NORMATIVE ARGUMENTS-

An analogy is a common way to support a normative statement, a claim that something should (or
should not) be done, or that it’s good (or bad) or morally right (or wrong). A general rule for normative
arguments is that a normative conclusion requires a normative premise. An analogical argument with
a normative main argument feature follows this rule.

Argument 8.3: ① It should be illegal for tanning salons to sell their services to minors
(people under 18). ② Just like smoking, tanning has proven health risks – it causes
cancer (skin cancer). Moreover ③ like smoking, tanning appeals to young people who
take their health for granted. ④ There is even evidence that people become addicted
to tanning, as they do to smoking. So in fact, ⑤ selling tanning services to minors is
really a lot like selling cigarettes to minors.

Main Argument Sub-argument

②
The analogy is explicit (⑤) but a premise is
⑤ ③
implicit: ⑥ “It should be illegal to sell
cigarettes to minors.” ⑥ ④

① ⑤

Does Argument 8.3 convince you that statement ① is true? You might criticize the argument by
disagreeing with any premise. Maybe you reject ⑥ because you think selling cigarettes to children
should be legal! A more reasonable criticism might be: ⑤ is a weak analogy. There are significant
differences between smoking and tanning. For example, there’s no way to smoke without cigarettes,
but it’s easy to tan without a tanning salon by just lying in sunlight.

- 189 -
 -UNIT 8 SKILLS-

You must be able to:

• Reconstruct an extended analogical argument.

• Recognize normative statements.

• Evaluate analogies and analogical arguments.

- 190 -
 -QUICK TEST QUESTIONS-

1) In an extended analogical argument, where is the analogy?

a) The conclusion of the main argument. c) A premise in the main argument and the
b) A premise in the sub-argument and the conclusion of the sub-argument.
conclusion of the main argument. d) A premise in the sub-argument.

2) What is a weak analogy?

a) The items are completely different. c) The items are different in any way.
b) The items are significantly different. d) The items don’t share the feature in the
main argument.

3) Which of the following is not a normative statement?

a) F is a good thing to do. c) F should not be done.
b) F is usually done. d) F is the right thing to do.

- 191 -
 -PRACTICE EXAMS-

PRACTICE EXAM 1

Write the implicit statement. Reconstruct the argument in standard form using statement numbers.
Evaluate the argument by discussing both of the main argument’s premises. Suggest a reasonable
criticism of at least one of them. This criticism may be your own personal opinion or something you
think a reasonable person might say.

① The Olympics are an outdated tradition that should not continue in the 21st century. ② Except
during the occasional pandemic or war, every two years the Olympics happen and it’s the same
embarrassing spectacle every time. ③ I always hope it’ll be the last one but then there’s another
one two years later. ④ I get that many people regard the Olympics as a noble and wonderful
event, but ⑤ they are basically just a giant international beauty pageant [competition]. ⑥ There
are competitors from all over the world who qualified to be there by winning national contests. ⑦
Both competitions have a big group performance with all the competitors present and then a
variety of competitive events. And ⑧ audiences and announcers sometimes are (or pretend to
be) interested in competitors’ personal stories but are mainly there to watch and judge the
physical competition.

- 192 -
 PRACTICE EXAM 2

Write the implicit statement. Reconstruct the argument in standard form using statement numbers.
Evaluate the argument by discussing both of the main argument’s premises. Suggest a reasonable
criticism of at least one of them. This criticism may be your own personal opinion or something you
think a reasonable person might say.

① Artificial intelligence is rapidly improving at the task of facial recognition, matching an image of
a face (e.g. from a security camera) to a face in a database. ② Now privacy supporters around
the world want laws banning corporations and police from using facial recognition technology. But
these people are misguided: ③ there’s nothing wrong with police use of facial recognition systems.
④ Privacy supporters who fear that we’ll all soon be living in a “surveillance society” where
everyone’s movements are tracked and recorded are just being paranoid.

Consider this: ⑤ a “super-recognizer” is someone with highly developed structures in the temporal
lobe of their brain that allow them to reliably identify far more faces than normal people can, and
⑥ police departments in the UK employ super-recognizers to identify people in security camera
images. Surely ⑦ there’s nothing wrong with police using super-recognizers. But ⑧ facial
recognition technology, much like the temporal lobe of a super-recognizer, is just a system that’s
highly effective for a function that normal people already do less well, although ⑨ both systems
make mistakes. ⑩ While it’s true that citizens don’t know about or consent to having their images
viewed by facial recognition technology, the same is true of super-recognizers.

- 193 -
 PRACTICE EXAM 3

Write the implicit statement. Reconstruct the argument in standard form using statement numbers.
Evaluate the argument by discussing both of the main argument’s premises. Suggest a reasonable
criticism of at least one of them. This criticism may be your own personal opinion or something you
think a reasonable person might say.

① Of the 6000-7000 distinct languages in the world today, about half are spoken only by small
communities, and several hundred are close to extinction, meaning that without active
preservation, they will soon disappear as their last speakers die. If you don’t see why you should
care about this, here’s a new way to look at it: while ② it’s not often noticed, ③ the extinction of
a language is actually a lot like the extinction of a biological species. ④ In both cases, the world
loses a wonderful kind of diversity that was created by a long evolutionary process. And as
everyone acknowledges, ⑤ it’s important to prevent the extinction of biological species.

And there’s more. ⑥ The scientific value of languages, based on all the things scientists can learn
about the evolutionary and other processes that created them, is similar to the scientific value of
biological species, although of course ⑦ scientists are also the first to acknowledge in both cases
that having the real thing existing in the world is much more wonderful than just having knowledge
about its past existence. ⑧ It’s not surprising that many linguists, as well as most speakers of
these endangered languages, want greater efforts to save them.

- 194 -
 PRACTICE EXAM 4

Write the implicit statement. Reconstruct the argument in standard form using statement numbers.
Evaluate the argument by discussing both of the main argument’s premises. Suggest a reasonable
criticism of at least one of them. This criticism may be your own personal opinion or something you
think a reasonable person might say.

① Allowing people to sell their personal data is a terrible idea, although ② that is currently a goal
of several companies and politicians. ③ They think that people should legally own – and be able
to sell – all the personal data that gets created about them in the medical system, social media,
and online shopping. ④ Some of this data is already being collected and sold in a huge data
marketplace of which few people are fully aware.

⑤ It might seem like a positive and empowering change until you reflect on the fact that ⑥ it
would be a terrible idea to allow people to sell their organs (e.g. kidneys) for transplant operations.
Notice that ⑦ your data and your organs are both created by you just in the course of living your
life. And crucially, ⑧ in both cases poor people would feel pressure to sell them if they had the
option of doing so – a marketplace would lead to the exploitation of people who are already
disadvantaged. I suspect that ⑨ some of these politicians are well-meaning but just haven’t
thought their plan through.


The argument in this exam is based on Jeong, Sarah. (2019 July 5). Selling your private information is a terrible idea. The
New York Times. https://www.nytimes.com/2019/07/05/opinion/health-data-property-privacy.html

- 195 -
 PRACTICE EXAM 5

Write the implicit statement. Reconstruct the argument in standard form using statement numbers.
Evaluate the argument by discussing both of the main argument’s premises. Suggest a reasonable
criticism of at least one of them. This criticism may be your own personal opinion or something you
think a reasonable person might say.

① Heroin addicts are at risk of deadly infectious diseases and overdoses. ② Society also spends
a lot of money putting these people in jail, either for possessing an illegal drug or for stealing
things to be able to buy it. ③ Many of these problems could be addressed if we learned to see
heroin addiction in a different way. Of course ④ we’d like to cure every heroin addict, but ⑤ it
should be legal for doctors to supply the opioid drug heroin to people with incurable addiction.

⑥ Here’s the idea: heroin addiction is really not much different from diabetes (a disease that
makes people unable to absorb sugar unless they take the drug insulin). ⑦ Just like diabetes,
heroin addiction can be a long-term, life-threatening, incurable condition. Moreover ⑧ like heroin
addicts who originally chose to use heroin, many adult diabetics get the disease partly from lifestyle
choices (e.g. poor diet). But ⑨ however they got their disease, the fact is that addicts and diabetics
can each live fairly normal lives with simple, cheap injections. ⑩ Maybe this perspective on heroin,
shocking though it is to many people, will be enough for society to approach the problem in a more
humane and effective way.

- 196 -
 -ANSWER KEY-

QUICK TEST
1) c 2) b 3) b

PRACTICE EXAM 1

Answer
Implicit statement: ⑨ Beauty pageants are an outdated tradition that should not continue in the 21 st
century.

Main argument Sub-argument

⑥
⑨ [B has Feature.] ⑦
④ [A is similar to B.] ⑧

③ [A has Feature.] ④

Evaluation [Example Answer]: This analogical argument is weak.

I think statement ⑨ is a fairly reasonable view to hold, although beauty pageants could be less sexist
than they used to be if they drop the swimsuit competition and emphasize women’s talents and
volunteer work more. Maybe something like beauty pageants should continue in the 21 st century. The
main problem with this argument is that the analogy, statement ④, is really weak. Despite the
similarities given here, the Olympics are not much like beauty pageants. Beauty pageants are about
who is most appealing to judges, but the Olympics are about elite performance in athletics and sports.
Most events don’t have judges, but even those that do (e.g. figure skating) have many physically
extremely difficult technical components. While the contestants in beauty pageants certainly practice
for them, this is nothing like the training to which Olympic athletes have devoted much of their lives,
spending countless hours improving the abilities.

Explanation (not required for exam answers)

The function of each statement in the main argument is given in [brackets]. This is not a required part
of the answer. The evaluation is an example answer. A good answer could express different opinions
of the argument or its premises, or suggest different criticisms of the analogy.

- 197 -
 PRACTICE EXAM 2

Answer
Implicit statement: ⑪ Facial recognition systems are similar to super-recognizers. (Also correct: Police
use of facial recognition systems is similar to police use of super-recognizers.)

Main argument Sub-argument

⑧
⑦ [B has Feature.] ⑨
⑪ [A is similar to B.] ⑩

③ [A has Feature.] ⑪

Evaluation [Example Answer]: This analogical argument is fairly strong.

I think ⑦ is true. Super-recognizers are just people who use their talent to prevent crime. It would
make no sense to insist that police departments hire only bad recognizers.
I think the analogy in ⑪ is strong enough to provide good support for the conclusion. However I can
see why some people might think there are important differences between facial recognition systems
and super-recognizers that are not acknowledged in this argument. For example:
▪ Super-recognizers are people who could decide whether to help police or corporations based on
whether doing that seems ethical, but computers can’t make those ethical decisions.
▪ Being recognized by a person is a normal and natural thing but being recognized by a computer
is weird and creepy.

Explanation (not required for exam answers)

The function of each statement in the main argument is given in [brackets]. This is not a required part
of the answer. The evaluation is an example answer. A good answer could express different opinions
of the argument or its premises, or suggest different criticisms of the analogy.

- 198 -
 PRACTICE EXAM 3

Answer
Implicit statement: ⑨ It’s important to prevent the extinction of languages.

Main argument Sub-argument

④
③ [A is similar to B.] ⑥
⑤ [B has Feature.] ⑦

⑨ [A has Feature.] ③

Evaluation [Example Answer]: This analogical argument is very weak.

I disagree with ⑤. While there may be some species that it’s important to save because they’re useful
to us, in general it doesn’t really matter whether a species continues or not. In fact in many cases
(e.g. mosquitos) it would probably be better if they did go extinct.
I also think that ③ is a weak analogy. There are important differences between languages and
biological species that are not acknowledged in this argument. For example:
▪ Languages can isolate people from each other culturally, socially, and economically, but biological
species don’t isolate people from each other.
▪ A biological species is a group of living creatures trying to survive, but the words and grammar
of a language aren’t living things and don’t try to survive. They are just a way in which people
choose to communicate with each other.
▪ While the disappearance of a biological species could have unpredictable effects on ecosystems,
it doesn’t seem that there are similar risks with the disappearance of a language.

Explanation (not required for exam answers)

The function of each statement in the main argument is given in [brackets]. This is not a required part
of the answer. The evaluation is an example answer. A good answer could express different opinions
of the argument or its premises, or suggest different criticisms of the analogy.

- 199 -
 PRACTICE EXAM 4

Answer
Implicit statement: ⑩ Selling one’s personal data is similar to selling one’s organs.

Main argument Sub-argument

⑥ [B has Feature.] ⑦
⑩ [A is similar to B.] ⑧

① [A has Feature.] ⑩

Evaluation [Example Answer]: This analogical argument is weak.

The analogy in ⑩ is surprising but actually pretty strong. It isn’t perfect. For example:
• Data is information and organs are parts of your body.
• You can make more data but you can’t regrow organs.
But I don’t think these differences are big enough or important enough to affect this argument.
However ⑥ seems totally wrong to me. I think people should be free to do what they want with their
bodies so long as it doesn’t hurt anyone else. Not only does selling an organ for transplant not hurt
anyone else, but it could even save someone’s life.

Explanation (not required for exam answers)

The function of each statement in the main argument is given in [brackets]. This is not a required part
of the answer. The evaluation is an example answer. A good answer could express different opinions
of the argument or its premises, or suggest different criticisms of the analogy.

- 200 -
 PRACTICE EXAM 5

Answer
Implicit statement: ⑪ It should be legal for doctors to supply insulin to diabetics.

Main argument Sub-argument

⑦
⑥ [A is similar to B.] ⑧
⑪ [B has Feature.] ⑨

⑤ [A has Feature.] ⑥

Evaluation [Example Answer]: This analogical argument is somewhat weak.

There can be no serious doubt about ⑪. It’s obviously true the diabetics should get insulin.
The analogy in ⑥ is fairly well supported in the argument but there are some important differences
between heroin and insulin that are not acknowledged. For example:
▪ Without insulin, a diabetic will die; without heroin, addicts will suffer terrible withdrawal symptoms,
but they won’t actually die.
▪ The two chemicals treat the two diseases in very different ways. Insulin is naturally made by the
body and a diabetic’s insulin shots simply replace it. But heroin is completely different from the
body’s own opioids (such as endorphins). It’s not naturally made by the body; it just happens to
attach to the same receptors in the brain.

Explanation (not required for exam answers)

The function of each statement in the main argument is given in [brackets]. This is not a required part
of the answer. The evaluation is an example answer. A good answer could express different opinions
of the argument or its premises, or suggest different criticisms of the analogy.

- 201 -
 ~9~
-GENERALIZING FROM A SAMPLE-

i. -POPULATION AND SAMPLE-

A sample is part of a larger population of items. Sampling is a work-saving procedure. We learn

about the population by measuring the sample and then generalizing from it to the whole population.

Generalizing from a Sample Argument Pattern

• N% of [Sample] has [Feature].

N% of [Population] has [Feature].

Argument 9.1: At Gotham General Hospital, 30% of

pregnancies with women aged 20-24 are unplanned, so
30% of pregnancies in women aged 20-24 in Gotham are
unplanned.

In Argument 9.1, the population is pregnancies in women aged 20-24 in Gotham, the sample is
pregnancies in women aged 20-24 at GGH, and the feature is unplanned.

If the feature is less or more common in the sample than it is in the population, a generalization based
on that sample will be an underestimate or overestimate.

Population
Sample Sample
Has
feature

Generalization will be: Generalization will be:

Underestimate Overestimate

- 202 -
ii. -SAMPLE SIZE-

Measuring a larger sample gives us a better ability to generalize. Sample size is indicated with n=.

Argument 9.2: At GGH, 30% of pregnancies with women aged 20-24 (n=500) are
unplanned, so 30% of pregnancies in women aged 20-24 in Gotham are unplanned.

Precision and Accuracy

People often use the words precision and accuracy as though they mean the same thing. In fact they
are completely different concepts. The precision of a statement is its exactness or specificity. The
accuracy of a statement is how close it is to the truth.

Statement 9.3: The speed of light is 155,286,903 m/s.

m
Statement 9.4: The speed of light is 271,306,682 /s.

Statement 9.5: The speed of light is around 300,000,000 m/s.

Statements 9.3 and 9.4 are equally precise. Statement 9.5 is less precise because of around.

The true speed of light is 299,792,458 m/s. That means that each statement is more accurate – closer
to the truth – than the previous one.

The concepts of precision and accuracy also apply with non-quantitative examples.

Statement 9.6: Stanley Park is in Canada.

Statement 9.7: Stanley Park is in BC, Canada.

Statement 9.8: Stanley Park is in Victoria, BC, Canada.

Each statement is more precise than the previous. The most precise, Statement 9.8, is inaccurate.

- 203 -
Archery is a helpful analogy. Generalizing is like shooting an arrow at a target. Precision is the target
size. Higher precision = smaller target. It’s more difficult to hit (be accurate with) a small target.

Margin of Error and Confidence Level

–pts % +pts

A complete statistical generalization includes a margin of error

(MoE), which quantifies the precision. The statistic (N%) is the
target’s centre and the MoE is the distance to either edge. The MoE
is given in the units of percentage, percentage points (pts).
Larger MoE = bigger target = lower precision.

Argument 9.9: At GGH, 30% of 26% 34%

pregnancies with women aged 20-24
(n=500) are unplanned, so 30% of
pregnancies in women aged 20-24 in
Gotham are unplanned (±4 pts).

MoE is calculated for the sample size at a chosen confidence level, the chance that the correct
percentage for the population is within this range. This is the strength of the argument (assuming an
unbiased sample – see Sec.iv). A 95% confidence level is common – if it’s not stated, as in Argument
9.9, we can assume 95%.+


MoE also depends on population size. In most of the examples in this unit, population size is the same: huge. This makes it
simpler to see the effect of different sample sizes. MoE also depends on the percentage measured in the sample. Often MoE is
simply given for a sample size, in which case MoE is calculated assuming a measurement of 50%. This is the maximum MoE.
For a measured percentage lower or higher than 50%, the MoE is smaller.
+
Although 95% is convenient for mathematical reasons, it is a choice. We could choose any confidence level. (And there are
other mathematically convenient but less memorable numbers, such as 68% and 99.7%.) A confidence level of 95% is standard
because it’s easy to remember and picture (a 19-in-20 chance), high enough to take the statistic seriously, but not so high that
the margin of error becomes huge.

- 204 -
With a small sample, the archer has bad aim (they’re shaky), so it’s hard to be accurate. To have high
confidence, they must shoot at a large target (low precision/large MoE).

If the target is smaller (higher precision/smaller MoE) but their aim is the same (same sample size),
they have lower confidence.

To have high confidence with a small target (high precision), they need better aim (larger sample).

- 205 -
The usefulness of a sample is determined by its size. Compare Argument 9.9 to 9.10:

Argument 9.10: At GGH, 30% of pregnancies in women aged 20-24 (n=500) are
unplanned, so 30% of pregnancies in women aged 20-24 in Gotham are unplanned (±5.3
pts; 99% Confidence).

The sample size (n=500) is the same, so it’s equally useful. But it’s been used differently. Argument
9.10 has a higher confidence level (99% vs. 95% in Argument 9.9) – it’s stronger. How is that possible,
if the higher confidence has not been earned by doing more sampling work? Higher confidence doesn’t
come for free! Here it’s been achieved simply by lowering precision (±5.3 pts vs. ±4 pts).

- 206 -
iii. -COMPARING STATISTICS-

Margin of error can be important when we compare statistics to find a difference.

Purchases from Argument 9.11: It seems that clothes bought

FashionPassion.com at night from our website FashionPassion.com
100%
are more likely to be returned. Last month’s
80%
Returned records (n=250) show that the return rate for
60%
nighttime purchases is 84% (±4.5 pts). But
40%
the records (n=350) show that the return rate
20%
for daytime purchases is only 32% (±4.1 pts).
0%
Nighttime Daytime

Argument 9.11 has two sub-arguments. Their conclusions support a final conclusion that asserts a
difference (“more likely”). Both statistics have MoE, shown with error bars in the chart. Since there’s
a large gap between error bars, this is strong evidence that there really is a difference between
nighttime and daytime purchases.

Purchases from Imagine instead that the measurements had been

FashionPassion.com
62% and 55%. These statistics are separated by
65% only 7 pts. Now the error bars overlap.
Returned
60%

55%

50%
Nighttime Daytime

This suggests there’s a difference, but we can’t be very sure. To be more sure of this small difference,
we need smaller MoE; to get smaller MoE (at the same confidence level), we need larger samples.

- 207 -
iv. -SAMPLE SELECTION-

Sample Bias

In a biased sample, the measured feature is more (or less) common than it is in the whole population
– but not just by bad luck (sampling error). Rather, it’s because the items in the sample have some
feature that makes them more (or less) likely to have the measured feature. To understand the sample
bias, we must know about this relationship between the measured feature and a sample feature. If
we know this background information, we know to expect an overestimate (or underestimate).

Here’s an analogy. We’re pouring hot water and cold water into a large bucket. To find the temperature
of the large bucket water (population), we measure the temperature of the hot water (sample).

Hot Cold Total Water (Large Bucket)

Temp
Overestimate

Hot Cold Total Water

Regardless of the amounts of hot and cold water we combine, the large bucket water temperature
must be between the hot and cold temperatures, which means that it’s lower than the hot water we
sampled. So measuring the hot water to find the large bucket temperature gives an overestimate.

Abby has been travelling around the world by airplane, visiting many cities. She thinks:

Argument 9.12: 60% of the cities I’ve

visited have a university (and I’ve visited
a lot of cities!). So about 60% of cities
have a university.

Can you see the problem here?

- 208 -
• This is a biased sample. Cities
• Background info: Universities
(measured feature) are % with
University
more common among cities Overestimate

(population items) with an

airport (sample feature).
• Probably the statistic is an With Airport Without Airport Cities
overestimate.

Convenience Samples vs. Random Samples

Sample bias is a risk when using a convenience sample, a sample composed of population items that
are available and easy to include. Convenience samples are not necessarily biased and are common in
many kinds of research. But we must be aware of the risk of bias and watch out for it.

Ideally we want to use a random sample (e.g. selected by lottery)

for which every item in the population has the same chance of
being selected. This sample could be randomly different from the
population, just by bad luck – there’s still the problem of sampling
error! But we have no reason to expect specifically an overestimate
(or underestimate).

- 209 -
 -UNIT 9 SKILLS-

You must be able to:

• Say how precision, MoE, confidence level, or sample size would change when one of the others

changes. (NOTE: You do not need to calculate MoE for this course.)

• Recognize and evaluate an argument in which sample size affects argument strength.

• Recognize a biased sample and explain what makes it biased.

- 210 -
 -QUICK TEST QUESTIONS-

1) A statistic that is less precise is more likely to be…

a) …true. c) …an overestimate.
b) …false. d) …an underestimate.

2) Margin of error…
a) …is smaller when the sample is smaller. c) …quantifies the precision of a statistic.
b) …is smaller when the confidence level is d) …quantifies the confidence level of a
higher. generalization.

3) Two statistical generalizations support a difference strongly when…

a) None of these. c) …the difference between them is small
b) …the high end of one MoE and the low end compared to their MoE.
of the other MoE include the same numbers. d) …their error bars on a bar chart overlap.

4) 10% of A-Ws (Ws with A) have B; 30% of Non-A-Ws have B. What percentage of Ws have B?
a) Not enough information to know c) Less than 10%.
b) More than 30%. d) Between 30% and 10%.

5) A sample is random if every item in the…

a) …population had the same chance of c) …sample is more (or less) likely to have
being selected for the sample. the feature being measured.
b) …sample had the same chance of being d) None of these.
selected for the population.

- 211 -
 -PRACTICE EXAMS-

PRACTICE EXAM 1

1) Which argument is stronger? Briefly explain your answer.

a. We tested water quality in buildings around the city (n=150), and 31% of them had water that
is too dirty to be safe to drink. So 31% of buildings in the city have dirty water (±7.4 pts).
b. We tested water quality in buildings around the city (n=300), and 31% of them had water that
is too dirty to be safe to drink. So 31% of buildings in the city have dirty water (±5.2 pts).
c. (a) and (b) are equally strong.

2) Answer the question based on the story. Explain your answer.

Bob and Abby work at sidewalk food carts downtown. Today Bob got the best (busiest) location,
where he will get more customers than Abby. Mostly they sell hot dogs, but about 10% of their
customers buy their deep-fried squid-on-a-stick. Today who is more likely to have at least ¼
of their customers order squid-on-a-stick: Abby, Bob, or are they equally likely?

3) Identify the elements of the argument and then evaluate it.

We learned that only 22% of houses in Gotham have smoke detectors with properly charged
batteries (±3.5 pts). We checked houses (n=550) in Burnside, the poorest neighbourhood in
Gotham and only 119 of them had smoke detectors with properly charged batteries.

a) State the population, sample feature, measured feature, sample size, margin of error, and
confidence level.
b) Evaluate the argument. Illustrate your evaluation with a bar chart.

- 212 -
 PRACTICE EXAM 2

1) Which argument is stronger? Briefly explain your answer.

a. We asked Hollywood producers (n=250) if they considered themselves bigshots, and 75%
did, so 75% of Hollywood producers consider themselves bigshots (±5.4 pts; 95%
Confidence).
b. We asked Hollywood producers (n=250) if they considered themselves bigshots, and 75%
did, so 75% of Hollywood producers consider themselves bigshots (±3.5 pts; 80%
Confidence).
c. (a) and (b) are equally strong.

2) Reconstruct and evaluate the argument.

① Gotham’s popular Deathtrap nightclub is not only the coolest club in town but also has the
highest success rate for getting patrons to buy the overpriced, watered-down drinks – greater
success than Gotham’s next most successful club, Candyland. ② 74% of Deathtrap patrons
buy the overpriced, watered-down drinks, based on the patrons we surveyed (n=54) recently
at Deathtrap. ③ 74% of our sample bought the drinks. But when we surveyed patrons (n=48)
at Candyland, ④ only 71% had bought the drinks, so ⑤ Candyland’s rate is only 71%.

3) Evaluate the reasoning of the news stories in the following (true) story.

Researchers at Boston University were interested in the link between (American) football and
Chronic Traumatic Encephalopathy (CTE), a neurodegenerative disease that causes cognitive
and mood problems. It’s believed to be caused by mild repetitive brain trauma of the sort that
football players normally experience even while wearing helmets. The researchers studied 202
football player brains that were donated to the CTE study by their families when they died.
They found CTE in nearly every brain. Many news headlines reported the findings by
announcing that almost all football players get CTE.

- 213 -
 PRACTICE EXAM 3

1) Which argument is stronger? Briefly explain your answer.

a. At our gadget factory, we tested gadgets coming off the assembly line to see if they work as
designed. In our sample (n=400), 88% work as designed. Therefore 88% of our assembled
gadgets work as designed (±3.2 pts).
b. At our gadget factory, we tested gadgets coming off the assembly line to see if they work as
designed. In our sample (n=400), 88% work as designed. Therefore 88% of our assembled
gadgets work as designed (±4.2 pts).
c. (a) and (b) are equally strong.

2) Evaluate the reasoning in the story.

Azmakia is having a referendum next week to decide whether to officially make the dung beetle
the national arthropod of Azmakia. Azmakian law requires 3-to-2 support for a referendum
proposal to pass. The dung beetle supporters have been campaigning for months and are now
sure they have enough support for success in the referendum. They asked 115 people from all
around Azmakia who plan to vote and 63% of them support the proposal.

3) Identify the elements of the argument and then evaluate it.

At the Wackadoodle gadget factory, the Quality Control Department needs to determine the
defective rate for the gadgets they produce. There are several employees in the factory making
gadgets, so QC picks one employee whose gadgets they will monitor. They choose Bob, a new
employee. It’s his first week on the job. They check Bob’s gadgets (n=320) and discover that
16% are defective. “Wackadoodle has a major quality control problem!”, they conclude. “16%
of our gadgets are defective! (±4 pts).”

a) State the population, measured feature, sample feature, sample size, margin of error, and
confidence level.
b) Evaluate the argument. Illustrate your evaluation with a bar chart.

- 214 -
 -ANSWER KEY-

QUICK TEST
1) a 2) c 3) a 4) d 5) a

PRACTICE EXAM 1

1) Answer
(c) – The arguments have the same (assumed) confidence level: 95%. The larger sample in
argument (b) is used to increase precision (smaller MoE) rather than increase confidence.

2) Answer
Abby. Her less busy location gives her a smaller sample of customers that is more likely to have
a different composition (at least ¼ who want squid-on-a-stick) than the population of all
customers (10% who want squid-on-a-stick).

3) Answer
a)
Population: Gotham smoke detectors [Also correct: Gotham houses]
Measured feature: has charged batteries [Also correct: has smoke detector with charged
batteries]
Sample feature: in Burnside
Sample size: 550 MoE: ±3.5 pts Confidence level: 95% (assumed)

b)
Evaluation: This argument is bad because it uses a biased sample. Burnside is a poor
neighbourhood where people have bigger problems to worry about than replacing smoke detector
batteries, and do little house maintenance. We can expect that charged batteries are less common
in Burnside Gotham smoke detectors. [Also correct: Burnside Gotham smoke detectors are less
likely to have charged batteries.] Probably 22% is an underestimate. The study should have
checked randomly selected houses from all over Gotham.

- 215 -
 Gotham Smoke Detectors
60%

40%

Example
Statistic
Has charged
batteries
20%

0%
Burnside Non-Burnside

Explanation
The population is a group of items, the subject of the conclusion. The measured and sample
features are not things – they are properties that population items may have or not have. The
margin of error is given in pts, not %. The confidence level is not the statistic, but the chance
that the true statistic is within the percentage range defined by the margin of error.

- 216 -
 PRACTICE EXAM 2

1) Answer
(a) – The arguments have samples of the same size but they do not use the sample in the same
way. Argument (b) uses the sample to increase precision (smaller MoE); argument (a) uses it to
increase confidence.

2) Answer
Main Argument Sub-argument Sub-argument

⑤
② ③ ④

① ② ⑤

Evaluation: This is a weak argument. The conclusion is based on a comparison of two statistical
generalizations – the drink-buying rates – that are based on fairly small samples of patrons at the
clubs. Using standard confidence levels, these statistics would have margins of error that are large
compared to the difference between them (3 pts). In other words, the margins of error would
overlap. Patrons at Deathtrap nightclub might be equally or even less likely to buy the drinks than
those at Candyland.

Explanation
The two sub-arguments each use the pattern (generalizing from a sample) introduced in this unit.
The main argument simply compares two statistics (its premises) and concludes that there is a
difference. This is the written version of the two-bar bar chart. In the chart, if margins of error
were known, they would be indicated with error bars.

3) Answer
The news stories treat the brains in the CTE study as a sample of football players’ brains. But this
sample is extremely biased. To be in the CTE study, someone’s brain had to be donated by their
family. Why would a family do that? In most cases, they did it because the person had cognitive
and mood problems, i.e. symptoms of CTE. So probably CTE (measured feature) was more
common among donated (sample feature) football player brains, and the statistic (“almost all”)
is an overestimate.

- 217 -
 PRACTICE EXAM 3

1) Answer
(b) – Although the confidence levels are unstated and this normally means 95%, it is impossible
for both arguments to have the same confidence level. The sample sizes are the same but the
statistic in argument (b) is less precise (larger MoE), so its confidence level must be higher.

2) Answer
Based on the measurement of their sample, they infer 63% support among planned voters, which
is high enough for the proposal to pass. However 63% is only 3 pts higher than the support
percentage required (3-to-2 = 60%). At any normal or useful confidence level (e.g. 95%), a
sample size of 115 has a margin of error larger than that. Since a failing support percentage
(59%) is well within the margin of error of this statistic, the campaign should not be so sure that
the dung beetle will be Azmakia’s new national arthropod.

3) Answer
a)
Population: Wackadoodle gadgets
Measured feature: defective
Sample feature: made by new employee
Sample size: 320 MoE: ±4 pts Confidence level: 95% (assumed)

b)
Evaluation: This argument is bad because it uses a biased sample. Bob is a new employee still
learning how to do his job well. Probably defects are more common among Bob’s gadgets. [Also
correct: Bob’s gadgets are more likely to be defective.] Probably 16% is an overestimate. The
study should have checked randomly selected gadgets from many different employees in the
factory.

Wackadoodle Gadgets
20%

15%
Defective
10%

5%
Example
Statistic
0%
Made by new Made by experienced
employee employee

- 218 -
Explanation
The population is a group of items, the subject of the conclusion. The measured and sample
features are not things – they are properties that population items may have or not have. The
margin of error is given in pts, not %. The confidence level is not the statistic, but the chance
that the true statistic is within the percentage range defined by the margin of error.

- 219 -
 ~ 10 ~
-CAUSE AND EFFECT-

i. -THE METHOD OF DIFFERENCE-

Causal Statements

Determining that one thing causes another is important both in science and in everyday life. A cause
produces an effect (makes it happen or exist). The effect happens after or together with the cause.

Many normal sentences can be restated as explicit causal statements.

Statement 10.1: Abby’s smartphone distracts her while

she is driving.
Statement 10.1 (explicit causal statement): Abby’s
smartphone causes her to be distracted while driving.

Many explanations are causal claims.

Statement 10.2: The dog barked because it saw a cat.

Statement 10.2: (explicit causal statement): Seeing
a cat caused the dog to bark.

Some causes are preventions. The effect they produce is a lack or reduction of something.

Statement 10.3: The HPV vaccine prevents cervical

cancer in women.
Statement 10.3 (explicit causal statement): The HPV
vaccine causes women to not get cervical cancer.


Vocabulary: The word cause is not the word because, but because often indicates an explanation, and many explanations
are causal statements. “A because B” often means “B caused A”.

- 220 -
 The Method of Difference

Causal reasoning often uses the method of difference.

Method of Difference Argument Pattern

Main Argument Sub-argument

• A is different when B is different. • C is the same when B is different.

• There are no other relevant differences • Etc.
besides A.
There are no other relevant differences
A causes B. besides A.

Argument 10.4: ① Normally Abby doesn’t text while driving, and she drives well,
but today she texted her friend and she crashed. ② She wasn’t driving in the rain for
the first time and ③ wasn’t trying to avoid hitting an animal on the road. So ④ texting
caused her to crash.

To reconstruct Argument 10.4,

Main Argument Sub-argument
we need to recognize an implicit
intermediate conclusion: ⑤ ① [A different when B different.] ②
“There were No Other Relevant ⑤ [NORD (besides A).] ③
Differences (besides texting)
④ [A caused B.] ⑤
when Abby crashed.”

Since the method of difference depends on (or works best when) finding the only relevant
difference, it can be criticized if we notice another difference that seems relevant (a possible cause).
What is relevant? That can only be judged based on background information about the effect.


There is also a method of agreement: A is the same when B is the same, therefore A causes B. This may be understood as a
method of searching for something we expect would be the only relevant difference between B situations and non-B situations.

- 221 -
ii. -CAUSE AND EFFECT-

To show a causal explanation (A causes B; B is an effect of A), we’ll draw a cause-and-effect diagram.

A B

To show prevention (A causes B to not happen), we’ll cross out B in the diagram.

A B

Multiple causes and effects may be ordered in a causal chain.

Distracted
while
driving

Most things have more than one cause and more than one effect.

Abby
distracted
while
driving

- 222 -
iii. -EXPLAINING DIFFERENCE-

A statistical difference: B is more common among W that have feature A. Why?

Three Causal Explanations of a Difference

A B
W [population]

B A B

C A Non-A

B Why is the B rate higher in the A group?

Argument 10.5: Security cameras are

less common in stores with lots of
Shoplifting
shoplifting, so shoplifting prevents
security cameras.

But maybe the true explanation is the reverse:

security cameras prevent shoplifting. Shoplifting

- 223 -
 Argument 10.6: Offices with ping pong
tables are more likely to have happy
employees, so ping pong tables make
employees happy.

But maybe there is a common (shared)

cause, something that separately causes
ping pong tables and happy employees.
For example, maybe nice bosses buy
ping pong tables for the office and make
Nice
employees happy by giving them raises.
Bosses $

We can easily eliminate the reverse causal explanation (B causes A) when B happens after A. Causes
happen before (or with) their effects, never after them.

In cases with no clear time order, and in cases where we think there may be a common cause,
eliminating the alternative explanations require an experiment.

- 224 -
iv. -CONTROLLED EXPERIMENTS-

A controlled experiment compares two groups: an experimental group and a comparison

group*. The independent (or manipulated) variable (A) should be the only relevant difference
between the two groups. We measure some feature, the dependent variable (B), in each group.

W
Dependent Variable

Independent Variable

A Non-A
Confounding Factor F Non-F

Confounding Factors and Alternative Explanations

The chart above shows that B is more common with A, and B is less common without A.

A A

B B

The confounding factor F is another difference (besides A) between the A and Non-A groups. It
could+ be a common cause of A and B.

A A

F F
B B

*
Vocabulary: A comparison group is often called a control group.
+
Vocabulary: In some scientific fields, confounding factor (or confound) = common cause. Here it means simply some other
difference besides the independent variable. It might be a common cause.

- 225 -
Or perhaps some other confounding factor, not yet discovered, is a common cause.

A A

? ?
F B F B

These explanation patterns are the same if B is less common with A, except that one thing is caused,
and the other thing is prevented, by some common cause.

A A

? ?
B B

Example

Philosophy book reading is more common among students who watch cat videos.

Students
50%

40% Reads
Philosophy
30% Books
20%

10%

0%
Watches Internet Does Not Watch
Cat Videos Internet Cat Videos

“Watching cat videos causes students to read

philosophy books” explains the higher philosophy-
reading rate among cat-video watchers. So the
difference is evidence for that causal relationship.

- 226 -
The difference is evidence, but not proof. Perhaps gender is a confounding factor.

Students
50%

40%
Reads
30% Philosophy
Books
20%

10%

0%
Watches Internet Cat Does Not Watch Internet
Videos Cat Videos

Male:Female ratio:
20:80 65:35

There is an alternative explanation: Being female is a common cause. Being female separately causes
both cat-video-watching and philosophy-reading. (And being male prevents both things.)

We need to control for gender: keep the male:female ratio the same in the cat video watcher
(experimental) group and non-cat video watcher (control) group. They could be all males, all females,
or any other ratio. Then we measure the rates of philosophy-reading again. Is there still a difference?

You’re doing an experiment…

PROBLEM F is a confounding factor. = F is another difference.

SOLUTION Control for F. = Keep F the same.

- 227 -
We control for gender: we compare female cat-video-watchers to female non-cat-video-watchers.

Possible Experiment Outcomes

The philosophy-reading rate is still higher in the The difference disappears. Gender matters.
cat video group. So gender doesn’t matter. This This supports “being female” as a common
eliminates “being female” as a common cause. cause. Why? Because if it’s true, we should
We now have better evidence than before that expect equal rates of the effect (philosophy-
watching cat videos causes philosophy reading. reading) when there are equal rates of the
cause (being female).

Difference Remains Difference Gone

Students Students
50% 50%
40% 40%
Reads
Reads
30% Philosophy 30%
Philosophy
Books
20% 20% Books

10% 10%
0% 0%
Watches Does Not Watch Watches Does Not Watch
Internet Cat Internet Cat Internet Cat Internet Cat
Videos Videos Videos Videos

Supports: original explanation Supports: alternative explanation

(Comparing males to males, we would see the

same low rates of philosophy reading instead.)

- 228 -
 Randomized Controlled Trials

We study 800 students: 300 who watch cat videos and

500 who don’t watch them.

Unknown to us, in our study population, 240 students

(30%) were dropped on their head as a baby: 225
(75%) of the 300 cat-video-watchers and 15 (3%) of
the 500 non-cat-video-watchers.

Students
50%
40%
30% Reads
Philosophy
20% Books
10%
0%
Watches Internet Does Not Watch
Cat Videos Internet Cat Videos

Dropped: 225 15
Group total: 300 500

Dropped rate: 75% 3%

There’s a higher “dropped” rate among the cat-video-watchers. Being dropped is a confounding factor.
Perhaps it’s a common cause of watching cat videos and reading philosophy, and that’s why cat-video-
watching students are more likely to read philosophy. (This would mean that not being dropped prevents
both things, although saying the explanation this way sounds a bit odd.)

- 229 -
Since we don’t know who was dropped on their head as a baby, we can’t control for this by directly
setting the dropped rate the same in each group. A randomized controlled trial (RCT) solves this
problem. Random = every student has 1) an equal chance of being assigned to the experimental
group and 2) an equal chance of being assigned the comparison group.

With coin-flip (50:50) randomization, we expect each group to get

half (400) of the total students (800), and half (120) of the dropped
students (240): the same 30% dropped rate in both groups.
Randomization automatically controls for being dropped.

Randomization might not produce exactly 30% and 30%, e.g. we

117 123
might randomly get /403 (29%) and /397 (31%). But we would
be very unlucky to randomly get rates as different as 75% and 3%.

Students (n=800, 240 dropped on head)

Reads
Philosophy
Books

Watches Internet Does Not Watch

Cat Videos Internet Cat Videos

Dropped: 120 120 Expected numbers with coin-flip

Group total: 400 400 (50:50) randomization

Dropped rate: 30% 30%

With adequate sample (group) sizes, we can trust randomization to automatically prevent confounding
factors. Now is philosophy-reading still more common with cat-video-watchers?


These two chances do not need to be equal to each other – as long as each is the same for every item in the population. So
then why coin-flip (50:50) randomization? Why not roll a dice ( 1/6 chance of assignment to the experimental group; 5/6 chance
of assignment to the control group)? Coin-flip randomization produces (roughly) equal group sizes. This is the optimal way to
minimize the margins of error on both statistics. That’s the only reason for the coin rather than the dice. If the sample (group)
sizes are large enough, we could eliminate confounding factors with dice-roll randomization instead.

- 230 -
 Possible Experiment Outcomes

Difference Remains Difference Gone

Students Students
50%

40%
Reads Reads
30% Philosophy Philosophy
20% Books Books
10%

0%
Watches Does Not Watches Does Not Watch
Internet Cat Watch Internet Internet Cat Internet Cat
Videos Cat Videos Videos Videos

No Confounding Factors No Confounding Factors

Supports: original explanation Supports: alternative explanation

(The rates would be the same, but what rate?

That would depend on the rate of the
(unknown) cause that is now equally present in
both groups.)

- 231 -
v. -NEW CONFOUNDING FACTORS-

An experiment may accidentally create a new confounding factor and a new alternative explanation
of a difference. In this type of explanation, the common cause may be the experiment itself.

Example

Cloud9 is an airline. Sometimes their airplanes break and need repair earlier than their normally
scheduled maintenance. Cloud9 considers upgrading their planes with the new JetBlast5000 engine.
They hope the JB5000 is more reliable and will prevent early repair need.

Before upgrading all their planes, they buy some JB5000 engines and do an experiment. To avoid
confounding factors, they randomly install JB5000s on some planes and regular engines on others.

Airplanes JB5000 planes need early repair less often. Cloud9

50% thinks that the JB5000 prevents early repair need.
40% Needs Early
30% Repairs

20%
10%
0%
JB5000 Regular
Engine
(Image: Angi Reinschmidt)

But the experiment may have created a new confounding factor. Perhaps Cloud9 told pilots of JB5000
planes to fly at high altitude because the engines work better there. But the air is less dense. Maybe
that prevents stress on the plane parts, preventing early repair needs. If this explanation is right, the
JB5000 does not prevent early repair needs.

EXPERIMENT

Stress on
plane parts?
High altitude flying

- 232 -
Randomization of the airplane groups did not control for altitude because the experiment itself (after
randomization) caused the JB5000 planes to fly at a higher altitude. But the problem can be fixed:
control for altitude. Now do JB5000 planes still need early repair less often?

Possible Experiment Outcomes

Difference Remains Difference Gone

Airplanes Airplanes
50% 50%
40% 40%
30% Needs Early 30% Needs Early
20% Repairs 20% Repairs
10% 10%
0% 0%
JB5000 Regular Engine JB5000 Regular Engine

100% 100% 100% 100%

Supports: original explanation Supports: alternative explanation

EXPERIMENT

High altitude
flying

Comparing 0% high altitude JB5000 planes to

0% high altitude regular planes, they would
see the same high rates of early repair need
instead.

- 233 -
 -UNIT 10 SKILLS-

You must be able to:

• Recognize causal statements.

• Recognize, reconstruct, and evaluate method of difference arguments.

• Identify or think of confounding factors in a story.

• Formulate causal explanations and draw cause-and-effect diagrams.

• Describe the randomization procedure of an RCT experiment.

- 234 -
 -QUICK TEST QUESTIONS-

1) What does “A prevents B” mean?

a) A does not cause B to happen. c) B causes A to happen.
b) A causes B to not happen. d) B causes A to not happen.

2) In the method of difference, we try to find the only relevant difference between…
a) …one situation when an effect happened c) …one possible cause of some effect and
and another when it did not. another possible cause of it.
b) …a cause and an effect. d) None of the above.

3) Which is possible? A causes B…

a) …also A separately causes C. c) …also C separately causes B.
b) …then B causes C. d) (a) – (c) are all possible.

4) Which does not explain a difference in the B-rate between A-Ws and non-A-Ws?
a) A and B separately cause C. c) A causes B.
b) B causes A. d) (a) – (c) all explain the difference.

5) What is a confounding factor?

a) Another difference between the c) Another difference between the controlled
independent and dependent variables. and uncontrolled experiments.
b) Another difference between the d) None of the above.
experimental and comparison groups.

6) What does it mean to control for F?

a) To measure F in a controlled experiment. c) To compare a group of F to a
b) To compare two groups between which comparison group of Non-F.
F is a difference. d) None of the above.

7) In a randomized controlled trial, items are randomly assigned to either…

a) …the randomized or the non- c) …the experimental or the comparison
randomized group. group.
b) …the cause group or the effect group. d) …the independent variable group or the
dependent variable group.

- 235 -
 -PRACTICE EXAMS-

PRACTICE EXAM 1

1) Write the implicit statement and give it a number. Reconstruct the argument in standard form
using the statement numbers. Evaluate the argument.

Near Gotham there are two crocodile petting zoos: Acme CrocWorld, and Big World o’Crocs.
Acme is near the hot springs, where geothermal heat warms the ponds, while Big World is in
Shady Valley, where there are no hot springs, but visitors enjoy a cool shady park while they
pet the crocodiles. Acme and Big World both breed crocodiles. Crocodiles lay eggs in the mud
around their waterholes. It’s normal for many eggs to never hatch, but the hatch rate at Big
World is significantly lower than at Acme. Big World’s Board of Directors calls in Zelda, the Zoo
Manager, and demands to know the problem. Zelda says:

① We’ve looked at the two petting zoos, and we think we know why our hatch rate is worse.
② Both zoos give the same amount of water and land space per crocodile, and ③ visitors are
kept away from the egg-laying areas where they might accidentally step on the eggs. However
our records shows that ④ Big World is a more popular petting zoo than Acme: every year we
get almost double their number of visitors, who all pet and bother the crocodiles. We think that
⑤ the constant attention from our extra visitors makes our crocodiles lay worse eggs that
hatch less often.

- 236 -
2) Answer the questions based on the story.

Cyclists in Gotham mostly use normal locks, but a new bike lock, SuperLock, is on the market.
The company does a survey asking cyclists what lock they use and if their bikes was stolen in
the past 6 months. SuperLock users had their bikes stolen more often (25%) than normal lock
users (15%)! Superlock had hoped their lock prevented bike theft but it seems to cause bike
theft instead!

a) Draw a bar chart of SuperLock’s survey data. Draw a cause-and-effect diagram of the
explanation they consider.

The company thinks of an alternative explanation for the shocking survey result: maybe
dedication to cycling causes some cyclists to buy the expensive SuperLocks, but dedication
also causes those cyclists to have their bikes stolen more often (the company can imagine
several different possible causal chains).

b) Draw a cause-and-effect diagram of the alternative explanation that SuperLock imagines.

Fill in the causal chain with your own suggestion.

The company will do an RCT experiment. They find 300 cyclists willing to participate.

c) Describe the experiment that the company will do. What are the dependent and
independent variables? Assume that 90 cyclists have the feature the company imagined
might be a confounding factor in the original survey.

The company has normal locks in November to begin the 6 month experiment. But the
SuperLocks have been sent to stores and they cannot start the experiment with those until
more are available in February. By August they have all the data. Once again the SuperLock
cyclists had their bikes stolen more often! They are depressed.

d) The RCT experiment outcome supports the original | alternative explanation.

e) Identify a new confounding factor the company created in their experiment. Draw a cause-
and-effect diagram of a new explanation of the difference. The diagram must include the
confounding factor you identified, plus one other that you think of.
f) How could the company have fixed the RCT experiment by changing the comparison group?

- 237 -
 PRACTICE EXAM 2

[This shorter exam combines the question formats of sections i and ii-v into a single question set.]

Write the implicit statement and give it a number. Reconstruct the argument in standard form using
the statement numbers. Then answer the questions.

The Azmakian military has been fighting in the country’s vast northern region against rebels
who want to overthrow the government. The government has a network of 40 outposts
throughout the northern region. The military’s job is to defend the outposts and chase any
rebels they can find. Recently the military has acquired some expensive new flying drones to
chase rebels. They have only 10 drones, though, so they use them at the outposts that are
surrounded by lots of steep mountains that would be dangerous for soldiers to guard on foot.

After a year, the military reviews data from all the outposts.
They say: “Look at this: ① the outposts protected by drones were attacked less often than
the other outposts (protected only by soldier guards). ② We should buy more drones! It seems
that ③ they were the only major difference between the outposts. ④ The drone-protected
outposts are just as strategically valuable as the other outposts, and ⑤ the fences and lights
we have around them are no better than we have at the other outposts. ⑥ The Azmakian
taxpayers won’t mind that these drones cost 35 million dollars each.”

a) Reconstruct the argument in standard form.

b) Draw a bar chart of the military’s data. (The exact numbers are not given and not necessary.)
Draw a cause-and-effect diagram of the explanation they have.
c) Evaluate their argument. Describe an alternative explanation of their data using details from the
story. Show your explanation by drawing in a cause-and-effect diagram.
d) The military didn’t think of the explanation you gave above but they want to do a better experiment
to be sure that the drones are truly effective before they spend the taxpayers’ money [Hahaha…].
Describe the experiment they should do. Suppose that 12 of the outposts have the feature you
mention in your answer to question (c). (Hint: Remember that they have only 10 drones.)
e) This year they do the better experiment that you described in your answer to question (d). The
drone-protected outposts get attacked less often. Does this fact support the military’s original
explanation or your alternative?
f) They look at the data and see that 40% of the drone-protected outposts were attacked while about
47% of the others were attacked. Does this information affect the answer to question (e)? Explain
why or why not.

- 238 -
 PRACTICE EXAM 3

1) Write the implicit statement and give it a number. Reconstruct the argument in standard form
using the statement numbers. Evaluate the argument.

To lower its high incarceration rate, Gotham City Council told the city’s two prisons to create
early release programs. Prison A releases every prisoner when they’ve served half their
sentence. Prison B was more concerned about endangering the community and wanted to do
more to stop ex-prisoners from committing new crimes. In their program, prisoners who’ve
volunteered for extra duties such as library attendant are released after half their sentence,
but they get a tattoo across their chest:

(backwards so it’s readable in a mirror)

After three years, Gotham City Council meets to discuss the programs.

“① The data here is pretty convincing. ② The DON’T COMMIT CRIMES tattoos are the only
significant difference between the two groups of early release prisoners. ③ There’s no
difference in the types of crimes they were convicted of, and ④ they’re placed in the same
types of job programs to help them re-enter society. Moreover ⑤ conditions in the prisons
(presence of gangs, treatment by guards, etc.) are the same. Yet ⑥ 40% of the prisoners
from Prison A commit another crime within a year of their release whereas only 10% of those
from Prison B do. ⑦ We’ll hire a tattoo artist for Prison A.”

- 239 -
2) Answer the questions based on the story.

Azmakia Broadcasting Network (ABN) makes comedy shows and is studying people’s reactions
to them so that they can make more funny shows. They show comedies to focus groups and
find a surprising difference between shows with and without a “laugh track” (added recorded
laughter): only 40% of the shows with a laugh track are rated “funny” by the focus groups,
whereas 75% of the shows without one are rated “funny”! ABN concludes that, contrary to the
TV industry’s belief, laugh tracks actually prevent shows from seeming funny.

a) Draw a bar chart of ABN’s data. Draw a cause-and-effect diagram of the explanation they
consider.

But ABN think of an alternative explanation. Perhaps good comedy writers, who write the
funniest jokes, also refuse to allow their shows to use laugh tracks (they think the laugh track
insults their talent as writers).

b) Draw a cause-and-effect diagram of the alternative explanation that ABN imagines.

ABN will study this with an RCT. They can edit any show to have or not have a laugh track.
They will study 42 shows.

c) Describe the experiment that the company will do. What are the dependent and
independent variables? Assume that 16 shows have the feature the company imagined
might be a confounding factor in the original survey.

The focus group for the shows with laugh tracks comes to the show testing centre and gives
their ratings. The next day the focus group for the shows without laugh tracks arrives after the
lunch break when the centre’s receptionist microwaved fish for her lunch. This time ABN finds
that shows with laugh tracks are rated funny more often, as the TV industry expects.

d) The RCT experiment outcome supports the original | alternative explanation.

e) Identify a new confounding factor the company created in their experiment. Draw a cause-
and-effect diagram of a new explanation of the difference. The diagram must include the
confounding factor you identified, plus one other that you think of.
f) How could the company have fixed the RCT experiment by changing the comparison group?

- 240 -
 PRACTICE EXAM 4

1) Answer the questions based on the story.

There was a heat wave across Azmakia last month. The Azmakian government collects data and
notices that across the country there were many reports of heat stroke (a serious medical
condition caused by overheating). But not every city had heat stroke cases. Some cities set up
public cooling stations (air conditioned rest places) during the heat wave. Few (a low percentage)
of the cities with cooling stations had heat stroke cases, whereas most (a high percentage) of the
cities without the stations had heat stroke cases. The government is hopeful that the cooling
stations protect people from heat stroke.

a) Draw a bar chart showing the government’s data.

b) Draw a cause-and-effect (arrows) diagram of the explanation that the government hopes is
true.

Although officials in the Azmakian Health Department are hopeful, they suspect a different
explanation of their data. Perhaps some cities have proactive city councils which organize the
cooling stations. And perhaps, for example, those councils also maintain lots of tree cover by
planting and protecting trees, and tree cover stops the heat from building up in urban areas.

c) Draw a cause-and-effect diagram of the alternative explanation of their data that the Health
Department thinks of.

Another heat wave is predicted next month and the Azmakian Health Department decides to
create a national cooling station program and run a national experiment to test the stations.

d) Describe how the government should do the experiment, and why they should do it this way.
Assume that they will include 30 cities in the experiment and 8 of them (they don’t know which
ones) have proactive city councils. Use these numbers in your explanation of the purpose of
doing the experiment in the way you describe.

e) They do the experiment and cities with cooling stations are less likely to have heat stroke
cases. This supports the original | alternative explanation.

- 241 -
 -ANSWER KEY-

QUICK TEST
1) b 2) a 3) d 4) a 5) b 6) d 7) c

PRACTICE EXAM 1

1) Answer
Implicit statement: ⑥ There are no other relevant differences between the zoos besides the
number of visitors.

Main Argument Sub-argument

④ [A different when B different.] ②

⑥ [No other relevant differences besides A.] ③

⑤ [A causes B.] ⑥

Evaluation: This is a bad method of difference argument. It depends on the claim (⑥) that the
number of visitors is the only relevant difference between the two zoos. But that statement is
false. The zoos are in different kinds of environments. Acme has hot springs but Big World does
not. This is another relevant difference, not considered in the argument. Perhaps Acme’s hot
springs cause the eggs to be healthier and have a higher hatch rate.

Explanation (not required for exam answers)

The role that each statement has in the method of difference main argument is given in the answer
above in [brackets]. This is not a required part of the answer.

2)
a) Answer

Cyclists
30%

Bike
20%
Stolen
Use Bike
10% SuperLock Stolen

0%
Use SuperLock Use Normal Lock

- 242 -
 Explanation
Other correct answers are possible for the chart. For example, the population could be Bikes,
the independent variable Locked with SuperLock (vs. Locked with Normal Lock), and the
dependent variable Stolen.

b) Answer

Uses Bike
SuperLock Stolen

Cycling Buys a More

Dedication Expensive Bike

Explanation
The possible confounding factor, Cycling Dedication, is a common cause in this explanation.
Another confounding factor, Buys a More Expensive Bike, is in the causal chain to the Bike
Stolen effect. This is an alternative explanation because there is no causal link between Use
SuperLock and Bike Stolen. This diagram refers to the experimental (Superlock) group. The
diagram for the control (Normal lock) group would be the same, but with a over each part.

Other correct answers are possible for the causal chain. For example, perhaps Cycling
Dedication causes people to keep their bikes shinier, or it causes them to take their bikes to
different areas where bike thieves go.

c) Answer
In their randomized controlled trial, the dependent variable is Bike Stolen and the independent
variable is Uses Superlock (vs. a normal lock). The company randomly (e.g. with a coin flip)
assigns each cyclist to either the experimental (SuperLock) group or the comparison (normal
lock) group. With 300 cyclists of which 90 are dedicated, we expect coin-flip randomization to
45
make each group /150 (30%) dedicated cyclists. Randomization automatically controls for
dedication.

d) Answer
The RCT experiment outcome supports the original explanation.

- 243 -
 Explanation
The SuperLock group still has a higher theft rate. Since randomization has (we assume)
eliminated confounding factors by controlling for dedication and other things, the SuperLocks
may be the only difference, and therefore the cause (however see the next answer!).

e) Answer
The experiment creates a new confounding factor: time of year. The comparison group (normal
locks) is in Winter; the experimental group (SuperLocks) is in Spring and Summer. The nicer
weather might cause people to take their bike out more and be targeted by thieves more.

Uses Bike
SuperLock Stolen

Experiment Cycles in Takes Bicycle

Spring/Summer Out More Often

Explanation
Other correct answers are possible for the causal chain. For example, perhaps demand for
stolen bicycles is higher in Spring and Summer, so thieves are more active then.

f) Answer
The company should run the comparison group later, during the same period as the
experimental group, to eliminate seasonal or weather-related confounding factors.

Explanation
Since they don’t have 150 SuperLocks available in November to run the experimental group in
Winter, delaying the comparison group is the only option.

- 244 -
 PRACTICE EXAM 2

a) Answer

Implicit statement: ⑦ The drones prevented attacks on the outposts.

Main Argument Sub-argument

① [A different when B different.] ④

③ [No other relevant differences besides A.] ⑤

⑦ [A prevents B.] ③

Explanation (not required for exam answers)

The role that each statement has in the method of difference main argument is given in the answer
above in [brackets]. This is not a required part of the answer.

b) Answer

Outposts

Attacked
Drone
Attacks
protection

Drone-protected Not Drone-protected

Explanation
The attack rates aren’t given so the bar chart simply shows a lower attack rate for drone-protected
outposts.

- 245 -
c) Answer
Statement ③ is false. There is another relevant difference between the outposts (a confounding
factor in the military’s experiment): the drone-protected outposts are surrounded by steep
mountains. The danger to soldiers of walking in these mountains is the reason why those outposts
were chosen for drone protection. But the mountains that are dangerous for soldiers would also
be dangerous for rebels and might stop rebels from attacking there. So the steep mountains might
be a common cause, causing the drones and preventing the attacks.

Drone
Protection Attacks

Steep Danger to
Mountains Rebels

d) Answer
They should do an RCT. Randomly assign each outpost to an experimental (drone) group or a
control (no-drone). Since there are 40 outposts and only 10 drones, they should give each outpost
a 1/4 chance (e.g. double heads, flipping two coins) of being assigned to the experimental group
– to choose 10. (Each outpost has a 3/4 chance of being assigned to the comparison group – to
choose 30.) Since there are 12 outposts with steep mountains, we expect 25:75 randomization
to makes groups with steep mountain rates: 3/10 (30%) and 9/30 (30%). Randomization
automatically controls for steep mountains.

e) Answer
The lower attack rate in experimental (drone) group supports the military’s original explanation.
This is good evidence that the drones are effective.

f) Answer
These percentages affect the answer to (e). Although the attack rate is lower in the experimental
group, the difference (7 pts) is fairly small. With the small sample (group) sizes (10 and 30), the
margins of error will be so large (at any normal confidence level) that they will significantly
overlap. These statistics provide only weak support for the drones’ effectiveness.

- 246 -
 PRACTICE EXAM 3

1) Answer
Implicit statement: ⑧ The tattoos prevent early released prisoners from committing new crimes.

Main Argument Sub-argument

③
⑥ [A different when B different.] ④
② [No other relevant differences besides A.] ⑤

⑧ [A causes B.] ②

Evaluation: This is a bad method of difference argument. It depends on the claim (②) that there
are no other relevant differences besides the tattoos, but that’s false. The prisoners in Prison B’s
program have volunteered for extra duties in the prison, whereas there’s no special requirements
in Prison A’s program. That difference is a relevant because it suggests that the Prison B prisoners
are especially motivated to reform and become goods citizen. Maybe that motivation, not the
tattoos, causes them to commit new crimes less often.

2)
a) Answer

Comedy Shows
100%

Rated
"Funny" Show
Laugh
50%
Track Rated
“Funny”

0%
Laugh Track No Laugh Track

Explanation
The cause-and-effect diagram shows the prevention explanation, referring to the shows in the
Laugh Track group. It’s important for question (b) to remember that there’s a corresponding
diagram for the No Laugh Track group (although it would be strange to say “No Laugh Track
causes shows to be Rated ‘Funny’”).
Laugh Show Rated
Track “Funny”

- 247 -
b) Answer

Laugh Rated
Track “Funny”

Good
Comedy
Funny Jokes
Writers

Explanation
See the explanation for the answer to question (a). The possible confounding factor, Good
Comedy Writers, is a common cause in this explanation. Another confounding factor, Funny
Jokes, is in the causal chain to the Rated “Funny” effect. This is an alternative explanation
because there is no causal link between No Laugh Track and Rated “Funny”. This diagram
refers to the control (No Laugh Track) group. The diagram for the experimental (Laugh Track)
group would be the same but with the switched to the other factors.

c) Answer
In their randomized controlled trial, the dependent variable is Rated “Funny” and the
independent variable is Laugh Track (vs. No Laugh Track). The company randomly (e.g. with
a coin flip) assigns each show to either the experimental (Laugh Track) group or the control
(No Laugh Track) group. With 42 comedy shows of which 16 have good writers, we expect
coin-flip randomization to make the Good Writer rate 8/21 (38%) in each group of shows.
Randomization automatically controls for good writers.

d) Answer
The RCT experiment outcome supports the alternative explanation.

Explanation
The difference in the Rated “Funny” rates reverse when (we assume) randomization eliminates
confounding factors. Laugh tracks now make a positive difference. This suggests that the
negative difference they originally found in the Rated “Funny” rates was caused by some
confounding factor (perhaps good comedy writers), not by the laugh track.

- 248 -
e) Answer
The experiment creates a new confounding factor: the microwaved fish. This will make the
show testing centre smell of the secretary’s leftover fish lunch while the comparison group (No
Laugh Track) watches the shows. The leftover fish smell might put the focus group in a bad
mood, which prevents them from rating the shows “Funny”.

Laugh Rated
Track “Funny”

Experiment Leftover Fish Focus Group

Smell Bad Mood

Explanation
The diagram refers to the shows in the comparison group.

f) Answer
The company should make sure the secretary doesn’t microwave her leftover fish lunch before
the focus group arrives!

- 249 -
 PRACTICE EXAM 4

1)
a) Answer

Cities

Had Heat
Stroke Heat
Cases Cooling
Stroke
Stations
Cases

Has Cooling Has No Cooling

Stations Stations

Explanation
The cause-and-effect diagram shows the prevention explanation, referring to the cities with
cooling stations. It’s important for question (b) to remember that there’s a corresponding
diagram for the cities with no cooling stations (although it sound strange to say “Lack of cooling
stations causes heat stroke cases”).
Cooling Heat Stroke
Stations Cases

b) Answer

Cooling Heat Stroke

Stations Cases

Proactive
City Extra Tree
Councils Cover?

Explanation
See the explanation for the answer to question (a). The possible confounding factor, Proactive
City Councils, is a common cause in this explanation. Another possible confounding factor,
Extra Tree Cover, is in the causal chain to the Fewer Heat Stroke Cases effect. This is an
alternative explanation because there is no causal link between Cooling Stations and Fewer
Heat Stroke Cases. This diagram refers to the cities with proactive city councils. The diagram
for the cities without proactive city councils would be the same but with the switched to
the other factors.

- 250 -
c) Answer
The government should do a randomized controlled trial with the cities in Azmakia. The
independent variable is Cooling Stations and the dependent variable heat stroke cases.
Assuming they have enough resources to set up cooling stations in 15 cities, they should do
50:50 randomization (e.g. with a coin flip). This is expected to assign 15 cities to a Cooling
Stations experimental group and 15 cities to a comparison group without cooling stations. If 8
of the cities have proactive city councils, we expect randomization to make two groups of cities
with rates of proactive city councils of 4/15 (27%). This means that they’ve automatically
controlled for proactive city councils (and for other possibly causal factors such as extra tree
cover).

d) Answer
The RCT experiment outcome supports the original explanation.

Explanation
The difference in heat stroke rates remains even after the government controls for proactive
city councils and other things that might have been confounding factors. This provides stronger
support for the original explanation. It’s likely that that the cooling stations prevent heat stroke
cases.

- 251 -