Unit 4 Tutorials: Practical Reasoning

INSIDE UNIT 4

Types of Inductive Reasoning

Inductive Reasoning
Causal Reasoning
Assessing the Strength of Causal Theories
Inference to the Best Explanation
Analogical Reasoning
Statistical Generalizations

Probability

Probability (Part 1)
Probability (Part 2)
Probability and Statistics
Errors in Reasoning about Probability

Moral Reasoning

Evaluative Language
Moral Truth (Cognitivism and Noncognitivism)
Moral Frameworks
Moral Dilemmas and Thought Experiments

Inductive Reasoning
by Sophia

 WHAT'S COVERED

In this tutorial, you will learn about different types of inductive reasoning and some governing principles
for engaging in inductive reasoning or analyzing inductive arguments. In particular, you will learn:

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 1
 1. Kinds of Inductive Reasoning
2. Invalidity Versus Weakness
3. Defeasibility

1. Kinds of Inductive Reasoning

 BEFORE YOU START

Recall that both deductive arguments and inductive arguments derive logical conclusions based on
premises, which are presumed to be true. The key difference is that deductive arguments can guarantee
their conclusion, and inductive arguments can only say that something is probably or usually true. Also
recall that the conclusion to an inductive argument should show that the argument is inductive, with a word
like "probably," or "likely," to show that the conclusion is not certain.
In Unit 2, we contrasted deductive arguments with inductive ones. We explained that while deductive
arguments have conclusions that must follow from the premises, an inductive argument has a conclusion that is
likely, or probably, true. The truth of an inductive conclusion isn’t absolute, for various possible reasons,
including:

The premises themselves are not absolute.

The argument is about a future event.
The argument is about a phenomenon that is too complex to sum up with a single declarative statement.

Induction is a broad area, encompassing specific kinds of reasoning that can be considered and evaluated
separately.

Causal reasoning: The branch of reasoning that explores the causes of known phenomena or the effects of
known causes. This is the reasoning most applicable to science and medicine.
Inference to the best explanation: The branch of reasoning that tries to explain how specific events came
to occur by appealing to the most likely explanation. This is the reasoning most applicable to forensics and
history and is also common in science.
Analogical reasoning: The branch of reasoning that discerns patterns and attempts to draw conclusions
based on comparisons. This is also commonly used in philosophy, as well as history, literary criticism, art
history, and other fields in the humanities.
Statistics: A branch of mathematics concerned with describing the likelihood of past (observed) events to
occur. Often it involves trying to describe the underlying causes of the observed data, and explain any
observed variation.
Probability: A branch of mathematics concerned with describing the likelihood of future (unobserved)
events to occur.

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 2
 This list is not comprehensive, but covers several ways you might encounter inductive reasoning in everyday
life. Inductive reasoning is crucially reasoning in the face of uncertainty. Statistics and probability are closely
related quantitative tools that can be used in making and strengthening inductive arguments.
We will further define each of these with examples and criteria for evaluating each kind of inductive argument.
However, they do have one thing in common: induction always involves making predictions about patterns in
one set of data you haven’t seen (or, generalizing) from another set of data. Induction is reasoning in the face of
uncertainty.

 TERMS TO KNOW

Causal Reasoning
The branch of inductive reasoning that explores the causes of known phenomena or the effects of
known causes.

Inference to the Best Explanation

The branch of inductive reasoning that tries to explain how specific events came to occur.

Analogical Reasoning
The branch of inductive reasoning that discerns patterns and attempts to draw conclusions based on
comparisons with other known phenomena.

Statistics
A branch of mathematics concerned with describing the likelihood of past (observed) events to occur.
Often it involves trying to describe the underlying causes of the observed data, and explain any
observed variation.

Probability
A branch of mathematics concerned with describing the likelihood of future (unobserved) events.

2. Invalidity Versus Weakness

Throughout Unit 3, we evaluated deductive arguments through truth tables and proofs. How do we evaluate
inductive arguments? In the truth table test, we simply find any rows where the premises are true and the
conclusion is false to determine that an argument is invalid. For an inductive argument, validity isn’t an
adequate measure.

Here’s why. Every deductive argument has a conclusion that is certain 100% of the time, and so will always be
inductively good. Suppose you have an argument whose conclusion has a 99% likelihood. While this argument
is inductively good, it’s deductively invalid because the conclusion is not guaranteed by the premises (that 1% is
killer). So many good, but not deductive, inductive arguments are technically invalid. Validity fails to distinguish
between good and bad inductive arguments.

Strength is the measure of quality for an inductive argument. A strong inductive argument is one where the
conclusion follows from the premises greater than 50% of the time; a weak inductive argument’s conclusion
follows 50% or less of the time.

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 3
 On the surface, inductive and deductive arguments look similar. They both have premises and conclusions, and
truth is the name of the game. Statements are written as declaratives. However, statements in inductive
arguments are probabilistic: they are true some proportion of the time, rather than absolutely all of the time.
Statements in deductive arguments are binary: they’re either true or false all of the time. There is no in-between
truth values for deductive arguments.

EXAMPLE In 2004, the Boston Red Sox played the New York Yankees in the American League
Champion Series. The Yankees won the first three games in the series. At this point, many people were sure
the Red Sox would lose the series. They had ample evidence supporting this position:

1. No team in the history of Major League Baseball had ever come back from a 0-3 deficit to win a 7-game
playoff series.
2. No team in the history of professional sports had ever done this. Basketball and hockey have similar 7-
game series throughout the playoffs, and no basketball or hockey team had ever won a series after
losing the first three games.
3. The Red Sox had infamously not won a championship in almost a hundred years and had lost a few in
dramatic fashion.
4. The Yankees, meanwhile, were a juggernaut team that had won several championships over the last
ten years.

In fact, as you may know, the Red Sox defied those odds and won the next four games, then went on to win
the World Series. So, the Yankees fan making an argument with those premises (or evidence) for the
conclusion that the Red Sox would lose, would be technically making an invalid argument. Yet, the
argument against the Sox was inductively strong; the Sox simply defied the odds. As the saying goes,
“That’s why they play the game.”
Even weak inductive arguments can turn out to have true conclusions. Surely some die-hard Sox fans in 2004
Boston were sure the Sox would come back and win the championship. Probabilistically speaking, the
likelihood that the Sox would win was small, but indeed non-zero; whereas, the likelihood that the Yankees
would win was much higher. The evidence that the Sox fans had was weak; the evidence that the Yankees fans
had was strong.

One of the most prominent differences between valid deductive arguments and strong inductive arguments is
the role that counterexamples play. A single counterexample to a deductive argument is sufficient to show
invalidity. On the other hand, a single counterexample to an inductive argument is not only fine but expected in
a strong inductive argument. Inductive arguments guarantee nothing, so some proportion of the outcome will
make the conclusion false. Remember, induction is reasoning in the face of uncertainty, but deduction
guarantees certainty.

 TERM TO KNOW

Probabilistic
Statements that are true some percentage of the time, as opposed to statements that are absolutely
true or absolutely false.

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 4
 3. Defeasibility
Inductive reasoning draws the best conclusion from the available evidence, but new evidence can change the
conclusion. Let’s look at another inductive argument:

Tweety is a healthy, normally functioning bird, and since most healthy, normally functioning birds fly, Tweety
can probably fly.

Notice that the conclusion, Tweety probably flies, contains the word “probably.” This is a clear indicator that the
argument is inductive, not deductive. Here is the argument in standard form:

1. Tweety is a healthy, normally functioning bird.

2. Most healthy, normally functioning birds fly.
3. Therefore, Tweety probably flies.

Given the information provided by the premises, the conclusion does seem to be well supported. That is, the
premises give us a strong reason for accepting the conclusion. This is true even though we can imagine a
scenario in which the premises are true and yet the conclusion is false. For example, suppose that we added
the following premise:

Tweety is 6 ft tall and can run 30 mph.

Were we to add this premise, the conclusion would no longer be supported. Any bird that is 6 ft tall and can run
30 mph is not a kind of bird that can fly. It must be an emu or ostrich.
This is called defeasibility. If an argument is defeasible, it means that new evidence could change the truth of
the conclusions drawn from the argument; it is capable of being falsified by new information or by
counterexamples. All inductive arguments (made in good faith) attempt the best or most likely conclusion, with
the understanding that if new information comes to light, the argument may need to be withdrawn.

EXAMPLE In the 1990s, several archeologists pursued a theory that Tibetan and Navajo people were
descended from the same ancient ancestors, presenting linguistic and cultural evidence showing
similarities, such as the use of sand mandalas in spiritual practice. This theory was consistent with the
widely agreed upon theory that the Americas were populated by people from Asia who crossed the Bering
Land Bridge between 10,000 and 20,000 years ago. While a reasonable inductive argument at the time,
and consistent with archeological theory and reasoning, the advent of DNA testing proved the theory
wrong, and any similarities were found to be coincidental. Moreover, new evidence placing humans in the
Americas earlier than originally thought raised challenges to the Bering Land Bridge theory. In both cases,
theories were supported by available evidence, but new evidence led researchers to reject the theories
they previously believed.
That fact that inductive arguments are defeasible does not mean that they are weak or poorly formed. It is
merely the nature of induction. Ideally, researchers should welcome new information that leads them to reject
faulty conclusions because the goal is to discover the truth, not to prove that their own theory is right.

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 5
 Scientific research is by nature inductive. Experimental studies aim to show that a test group (which receives an
experimental manipulation or treatment) has better outcomes than the control group (which does not receive
the manipulation or treatment). This comparison assumes that effects seen in the test group can generalize to
the control group, had they received the same manipulation/treatment. This process is by nature defeasible. If
the effect size is too small, the results might not be replicable. If the experimental sample is not representative,
the findings won’t generalize to the broader population. Researchers can make errors in the experimental
design or analysis. The favored explanation for the result might not be distinguishable from an alternative
explanation. All these possibilities can make the conclusion less reliable, so researchers strive to control these
confounds as much as possible.

Defeasibility is actually crucial to the scientific process. We observe some data, we construct a hypothesis to
explain why the data occurred, and we develop experiments that can tell us whether our hypothesis is right.
Scientific hypotheses must be defeasible, meaning it must be possible to prove them either right or wrong.
Otherwise, we learn nothing about the world.

The policy of the Open Science Framework is to make research design and analysis information widely
available to facilitate bringing to light any information that might defeat the results. To reiterate, ideally the aim
of science is not merely to prove that a particular researcher is correct, but to advance our collective knowledge
(and maybe even better our lives).

Unfortunately, this principle of defeasibility is frequently contaminated by personal interest. Prosecutors often
try to prevent introducing DNA evidence that could prove a suspect is innocent, because success in their career
is measured by conviction rates, and some value career success more highly than justice. A scientist who has
pursued a theory for decades might continue to do so despite substantial new evidence that bears against it.
These are manifestations of egocentric bias, and possibly also the sunk cost fallacy discussed in an earlier
tutorial. As we engage in inductive reasoning, we should seek to recognize what is to be learned even in the
cases where we were proven wrong. Errors too provide information that takes us closer to truth. If we reason in
good faith and are open about our assumptions and the evidence we had available at the time of reasoning,
then there’s no shame in being wrong in the face of new evidence.

Now that we understand the qualities of inductive arguments and how to approach them, we’ll take a closer
look at each of the five types we identified earlier.

 TERM TO KNOW

Defeasibility
In inductive reasoning, the quality of an argument where new evidence might change our conclusion.

 SUMMARY

In this lesson, you began to learn about identifying and evaluating kinds of inductive arguments,
particularly the notion of strength and weakness versus validity and invalidity. You learned how
statements of inductive arguments are probabilistic instead of true or false. You learned about the
principle of defeasibility, meaning that additional evidence may change our conclusion. Being

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 6
 defeasible does not mean an inductive argument is weak; rather, it is an inherent condition of inductive
arguments.

Source: THIS CONTENT HAS BEEN ADAPTED FROM Introduction to Logic and Critical Thinking.

 TERMS TO KNOW

Analogical Reasoning
The branch of inductive reasoning that discerns patterns and attempts to draw conclusions based on
comparisons with other known phenomena.

Causal Reasoning
The branch of inductive reasoning that explores the causes of known phenomena or the effects of known
causes.

Defeasibility
In inductive reasoning, the quality of an argument where new evidence might change our conclusion.

Inference to the Best Explanation

The branch of inductive reasoning that tries to explain how specific events came to occur.

Probabilistic
Statements that are true some percentage of the time, as opposed to statements that are absolutely true
or absolutely false.

Probability
A branch of mathematics concerned with describing the likelihood of future (unobserved) events.

Statistics
A branch of mathematics concerned with describing the likelihood of past (observed) events to occur.
Often it involves trying to describe the underlying causes of the observed data, and explain any observed
variation

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 7
 Causal Reasoning
by Sophia

 WHAT'S COVERED

In this lesson, you will learn about causal reasoning, or the events that result in other events.
Specifically, this lesson will cover:
1. Causation
2. Determining Causes
3. Determining Effects

1. Causation
Recall that in an earlier tutorial, we talked about illusory causation, the (incorrect) feeling that one event has
caused another. For example, the perceived effectiveness of a homeopathic cold remedy could actually be a
result of the body’s natural immune response. The causal link between recovery and the homeopathic
treatment may be completely imagined.

Crucial to this concept is causation, the determination that one event causes another to occur. True causation is
not illusory: if you throw a rock, you cause it to fall. If you strike a match, you cause it to ignite.

But causation is more complicated than that. Let’s take the example of striking a match. What if the matchbook
is wet? Or you’re in an oxygen-free vacuum (as in outer space)? If either of those situations is the case, then
striking the match will not cause a flame. So, it isn’t simply the action of striking the match that causes the flame,
but a combination of the action with a number of other conditions that must hold: the match head must
generate enough friction against the striking surface for the phosphorus to ignite, and there must be oxygen
present for the chemical reaction to occur. Which of those conditions we call the “cause” depends in part on the
context.

Suppose that you are striking a match in outer space (you, of course, are in a space suit with oxygen, but are
striking the match in the vacuum of space). You continuously strike the match, but no flame appears. One of
your fellow astronauts brings out a can of compressed oxygen and sprays it on the match while you strike. All of
a sudden, a flame is produced. In this context, the sudden expulsion of oxygen seems to “cause” the fire.

In the real world, the cause for any occurrence is usually more complex than one event. Rather, there are
usually multiple conditions that must be in place for any cause to occur. These are called background
conditions. We often take for granted the background conditions in normal contexts and just refer to one
particular event as the cause. In the case of striking a flame, these include (in most cases) the presence of

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 8
 oxygen and the absence of water or anything else that would prevent the high friction needed to ignite the
match. This isn’t an error, but a mere convenience.

For just about any cause, there are a number of conditions that must be in place in order for the effect to occur.
These are called necessary conditions. A necessary condition must occur to make the event happen, and as
noted, there are often multiple necessary conditions in play, some we don’t think about unless something goes
wrong.

A necessary condition of the match lighting is that there is oxygen present. A necessary condition of a car
running is that there is gas in the tank. We can use necessary conditions to diagnose what has gone wrong in
cases of malfunction. That is, we can consider each condition in turn in order to determine what caused the
malfunction. If the match doesn’t light, we can check to see whether the matches or striking surface are wet (or
have been wet and dried out, which we might not have noticed, and which would prevent us from achieving the
chemical reaction needed to ignite a flame). In contrast, a sufficient condition is one which, if present, will
always bring about the effect. For example, although striking a match is one of a few necessary conditions to
make it ignite, dropping the matchbook in water is sufficient to make it fail with or without other conditions.

Because the natural world functions in accordance with natural laws (such as the laws of physics), causes can
be generalized. For example, any object near the surface of the earth will fall towards the earth at 9.8 meters
per second unless impeded by some contrary force, such as a net or a strong gust of air, if the object is
lightweight. This generalizes to apples, rocks, people, and every other object.

Causal generalizations are often parts of explanations. For example, we can explain why the airplane crashed
to the ground by citing the causal generalization that all unsupported objects fall to the ground and by noting
that the airplane had lost any method of propelling itself because the engines had died. So, we invoke a causal
generalization in explaining why the airplane crashed.

Causal generalizations have a particular form:

For any x, if x has the feature(s) F, then x has the feature G.

For example: For any match (x), if that match head achieves sufficient friction while exposed to oxygen (F), it will
ignite (has the feature G). Or, for any match (x), if submerged in water (F), then the match will fail to light (has the
feature G).

 TERMS TO KNOW

Causation
The determination that one event causes another to occur.

Background Conditions
The totality of things, many assumed, that must be in place for an event to occur.

Necessary Conditions (in Causality)

Factors that must be in place for an event to occur.

Sufficient Conditions (in Causality)

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 9
 Factors that are enough by themselves for an event to occur.

Causal Generalization
A statement about the cause of a phenomenon, that generalizes beyond the specific observed event.
They have the form: for any x, if x has the feature(s) F, then x has the feature G.

2. Determining Causes

Being able to determine when causal generalizations are true is an important part of becoming a critical thinker.
Since in both scientific and everyday contexts, we rely on causal generalizations to explain and understand our
world, the ability to assess when a causal generalization is true is an important skill.

For example, suppose a veterinarian is trying to figure out what causes a dog, Charlie, to have seizures. The vet
will have a lot of questions that may isolate potential causes. Since Charlie doesn’t have a history of seizures,
the vet will look for recent changes. Did Charlie eat any human food? Did his people use a new shampoo or
bug repellent? Has he come in contact with other dogs? Is there any combination of those things that trigger
the seizures? If a cause can’t be found immediately, they may ask his people to keep a log of when the seizures
occur, and note the presence or absence of potential causes.

Day Seizure Human Food Shampoo Bug Repellent Dog Park

1 Yes No Yes Yes Yes

2 No Yes Yes No Yes

3 No Yes No Yes Yes

4 Yes Yes Yes Yes No

5 No Yes No Yes Yes

6 No Yes No Yes Yes

7 Yes Yes Yes Yes Yes

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 10
 A log like this has a target, the event we are trying to figure out, and candidates or potential causes. In this
case, the target is a seizure, and the candidates are eating food not designed for dogs, the shampoo used in
Charlie’s baths, bug repellent, and any environmental factors at the dog park (which may be other dogs,
herbicide used on the grass, or other features we can’t separate out).

 BIG IDEA

In practice, such logs are common in treating conditions for both humans and animals, and may be used for
other purposes as well. For example, if a store experiences occasional theft, they might keep track of the
workers, the day, any events outside the store which might affect their clientele, etc. If your car keeps
breaking down, you might track the weather, the kind of gas and oil, your speed, etc. “Keeping track” is
itself a good critical thinking skill, but also requires thoughtful judgment about what needs to be tracked.
The first thing we’d want to know is what candidate is present every time the target is true. This is called the
necessary condition test. Any candidate that is present every time Charlie has a seizure is a candidate for a
necessary condition. According to the data in the log above, we see that Charlie only has seizures on days he is
exposed to particular things, and that can tell us something about the cause of the seizures. We can also
eliminate any features that are not present on the days he has seizures. For example, human food is not
present on the first day he has a seizure, so we can (luckily for Charlie) remove that as a candidate for a
necessary condition. We can also remove going to the dog park since it did not occur on day 4, when he did
have a seizure. This leaves bug repellent and baths as two (likely) candidates for necessary conditions.

The next thing we’d want to know is if Charlie has a seizure every time a certain feature is present. This would
be a sufficient condition test and would reveal whether one candidate alone causes seizures. In other words,
any candidate that is present when the target is false is ruled out as a sufficient condition. Notice that there are
days when Charlie either got a bath or was treated with bug repellent, but he did not have a seizure on those
days. This means we can eliminate them as sufficient conditions (enough alone to cause a seizure), but not as
necessary conditions (present when he does have a seizure).

Although no one feature is sufficient for causing seizures (according to the data we observed), it is still possible
that certain features are jointly sufficient. Two candidate features are jointly sufficient if there is no case in
which both candidates are present and yet the target is absent. Applying this test, we can see that the shampoo
and bug repellent are jointly sufficient for the target feature since any time both are present, the target feature
is always present. Thus, from our data, we can infer that the likely cause of Charlie’s seizures are when we both
give him a bath and then follow that bath up with a bug repellent. Every time those two things occur, he has a
seizure (sufficient condition); and every time he has a seizure, those two things occur (necessary condition).
Thus, the data gathered so far supports the following causal conditional: Any time Charlie is given a shampoo
bath and a flea treatment, he has a seizure.

 TRY IT

How would you express these conditions using sentential logic?

Although in the above case, the necessary and sufficient conditions were the same, this needn’t always be the
case. Sometimes sufficient conditions are not necessary conditions. For example, while the shampoo and bug
repellent are jointly sufficient to cause Charlie to have a seizure, they may not be the only chemicals to do so. In
any case, determining necessary and sufficient conditions is a key part of determining a cause.

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 11
 This is a simplified example. We have assumed that Charlie’s reaction will occur on the same day, and that
environmental triggers are either completely true or completely false. We also have a quite short list of
candidates, whereas in a real health study there might be dozens of candidates. But it should be clear that
keeping a log like this will give you valuable data that can be analyzed to determine causes and is a great
critical thinking tool.

 TERMS TO KNOW

Target
The event we are trying to determine the causes for in testing for causality.

Candidates
In testing for causality, the factors that may be (alone or in part) a cause for the target outcome. These
are the factors that are tracked in a log.

Necessary Condition Test

A method for identifying a necessary condition on a log tracking background conditions and target
outcomes. If the target outcome never occurs when the candidate is absent, the candidate is likely to
be a necessary condition.

Sufficient Condition Test

A method for identifying a sufficient condition on a log tracking background conditions and target
outcomes. If there is always an outcome when a candidate is present, it is likely to be a sufficient
condition.

Jointly Sufficient
A kind of sufficient condition where there is always a target outcome when two or more background
conditions are all present.

3. Determining Effects
In this example above, we showed one side of causal reasoning. The other side is testing the possible effects of
something like a new product or medication. Since chemicals affect people and animals in different ways, this
isn’t something that can be determined by a single testing event. Further, because effects can be delayed, or
only due to long-term use of a product or medication, the testing cannot be done quickly.

Determining effects is similarly rigorous and detailed but is forward- rather than backward-facing. For example,
the introduction of a new medicine to the public is typically preceded by a long-term study in which a control
group (who do not get the drug) and a test group (who do) are tracked and compared. (Even prior to this stage,
the medicine undergoes animal testing trials to determine whether it is safe for humans.) After both test and
control groups are given their treatments, scientists carefully measure and compare the effects to determine
whether the two groups differ significantly from each other. In the best case, the test group will show significant
improvement while the control group does not change. These differences are measured using statistics. The
stronger an effect is statistically, the more we can believe that the medication is working.

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 12
  SUMMARY

In this tutorial, you learned about causal reasoning, the inductive process that analyzes causation, or
the relationships between cause and effect. This might be done for determining causes, such as
diagnosing allergies, or determining effects, such as medical testing. In determining causes, we might
attempt to control exposure to see how effects vary with or without specific potential causes and must
also remember to see if there are a combination of things that are the cause. In determining effects, the
ideal process is to have a test group and a control group so we can compare outcomes. Either way, the
principle of causal reasoning is the careful collection of data. This requires thinking broadly about
potential causes that need to be tested, and a commitment to comprehensive and accurate
information.

Source: THIS CONTENT HAS BEEN ADAPTED FROM Introduction to Logic and Critical Thinking.

 TERMS TO KNOW

Background Conditions
The totality of things, many assumed, that must be in place for an event to occur.

Candidates
In testing for causality, the factors that may be (alone or in part) a cause for the target outcome. These are
the factors that are tracked in a log.

Causal Generalization
A statement about the cause of a phenomenon, that generalizes beyond the specific observed event.
They have the form: for any x, if x has the feature(s) F, then x has the feature G.

Causation
The determination that one event causes another to occur.

Jointly Sufficient
A kind of sufficient condition where there is always a target outcome when two or more background
conditions are all present.

Necessary Condition Test

A method for identifying a necessary condition on a log tracking background conditions and target
outcomes. If the target outcome never occurs when the candidate is absent, the candidate is likely to be a
necessary condition.

Necessary Conditions (in Causality)

Factors that must be in place for an event to occur.

Sufficient Condition Test

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 13
 A method for identifying a sufficient condition on a log tracking background conditions and target
outcomes. If there is always an outcome when a candidate is present, it is likely to be a sufficient
condition.

Sufficient Conditions (in Causality)

Factors that are enough by themselves for an event to occur.

Target
The event we are trying to determine the causes for in testing for causality.

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 14
 Assessing the Strength of Causal Theories
by Sophia

 WHAT'S COVERED

In this lesson, you will learn more about the process for testing and validating causal theories.
Specifically, this lesson will cover:
1. Rigorous Testing
2. Concomitant Variation
3. Accidental Correlation

1. Rigorous Testing
In the previous tutorial, we saw an example using data to find potential causes for something (in this case, a
dog’s illness). However, it is important not to make hasty generalizations from this data. While this method is a
good way to identify likely causes for something, or narrow down the list of possible causes, it should be
followed up with rigorous testing. Rigorous experimental testing allows us to make stronger causal inferences.

Here is an example of rigorous testing. Recall that our dog Charlie’s veterinarian had asked us to keep track of
what Charlie was exposed to each day, and whether he had a seizure. Suppose that each day we collected
data, Charlie ate human food. Suppose that on none of the days, he was given a shampoo. Our data would now
look like this (note the differences from before):

Day Seizure Human Food Shampoo Bug Repellent Dog Park

1 No Yes No Yes Yes

2 Yes Yes No No Yes

3 No Yes No Yes Yes

4 No Yes No Yes No

5 No Yes No Yes Yes

6 No Yes No Yes Yes

7 Yes Yes No Yes Yes

Given this data, “Human Food” looks like a necessary condition because it is always present when Charlie has a
seizure. However, in order to rigorously test whether these table scraps are a necessary condition, we have to

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 15
 look for cases in which it is not present and then see whether our target condition is present. We can’t do this if
he has human food every day! We may as well say his fur coat is a necessary condition.

Similarly, shampoo appears ruled out as a sufficient condition because it is never present on days Charlie has a
seizure. However, to rigorously test shampoo as a sufficient condition, we have to look for cases in which
Charlie did have a shampoo and did not have a seizure. Since we haven’t given him a bath, we really don’t
know whether it’s a sufficient condition.

In rigorous testing, we are actively looking for (or trying to create) situations in which a candidate feature fails
one of the tests.

 HINT

This is the same idea as the concept of defeasibility that we covered in the previous challenge. To prove
that something is true, we need to give it a test it can fail.
Necessary Condition Sufficient Condition

Must be present for the target Will cause target event to occur
Meaning
event to occur. with no other conditions.

Seek out cases where the Seek out cases where the
To test with rigor
candidate is false. candidate is true.

Find a case where the candidate is Find a case where the candidate is
To discard
false and the target is true. true and the target is false.

If we are interested in causes, we should always rigorously test each candidate. This means that we must have
a mix of different situations where the candidates and targets are sometimes present and sometimes absent.
For example, we might remove all elements from Charlie’s life and reintroduce them one at a time to see if and
when the seizures recur. Or we can do it the opposite way, removing one item at a time completely, and seeing
if the problem goes away. This gives us the best data to analyze and the most insight into potential causes.

2. Concomitant Variation
The necessary and sufficient conditions tests can be applied when features of the environment are wholly
present or wholly absent. However, in situations where features of the environment are always present to some
degree, these tests will not work (since there will never be cases where the features are absent; therefore,
rigorous testing cannot be applied).

For example, suppose we want to know whether is a contributing cause to higher global temperatures. In
this case, we can’t very well look for cases in which is present but high global temperatures are absent
(sufficient condition test), since and high temperatures are always present to some degree (so to speak).
Nor can we look for cases in which is absent when high global temperatures are present (necessary
condition test), since, again, and high global temperatures are always present to some degree.

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 16
 Rather, we must use a different method, the method that John Stuart Mill called the method of concomitant
variation. In concomitant variation, we look for how things vary in accordance with one another. For example, if
we see that as levels rise, global temperatures also rise, then this is evidence that and higher
temperatures are positively correlated. When two things are positively correlated, that means that as one thing
increases, the other also increases at a similar rate (or as one decreases, the other decreases at a similar rate).

In contrast, when two things are negatively correlated, that means that as one increases, the other decreases
at a similar rate (or vice versa). For example, if a police department increased the number of police officers on
the street, and the number of crimes reported decreases, then the number of police on the street and number
of crimes reported would be negatively correlated. In each of these examples, we may think we can directly
infer the cause from the correlation—the rising levels are causing the rising global temperatures, and the
increasing number of police on the street is causing the crime rate to drop.

 TERMS TO KNOW

Concomitant Variation
The measure of two factors over time to see how one relates to the other.

Positive Correlation
When measuring two factors shows that both roughly increase and decrease at the same time.

Negative Correlation
When measuring two factors shows that one increases as the other decreases and vice versa.

3. Accidental Correlation
However, we cannot directly infer causation from correlation. As you might remember from earlier, illusory
causation is when we presume that because two things are correlated, like ice cream sales and drownings, that
one causes the other. Of course, both ice cream sales and drownings go up when the weather is hot and go
down when the weather is cold, and so might be correlated without causation at either end—this can be
explained because people swim more when it is hot out, and thus are at greater risk of drowning, and also
people eat more ice cream when it's hot out.

In other cases, there may be evident correlation but no explanatory theory of how one can cause the other or
how both can be affected by a third factor. This is called accidental correlation. Two correlated variables are not
amenable to a causal explanation. There are two types of accidental correlations: nonsense and spurious.
Nonsense correlations have no proper causal interpretation; it’s meaningless. Spurious correlations often have
a causal meaning outside the bounds of the two variables in question; there’s usually an unknown third variable
that accounts for the correlation between the other two.

For instance, the rate of drownings in swimming pools happens to be correlated with the release of films
starring Nicolas Cage. There is no way to explain this correlation—Nicolas Cage movies have not been found to
inspire swimming pool drownings, and there is no third factor that could be causing both pool drownings and
Nicolas Cage movies. This is a nonsense correlation. It doesn't mean anything; it is a coincidence.

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 17
 Sometimes two things can be correlated without either one causing the other. Rather, some third thing causes
them both, like hot weather causing ice cream sales and drownings to go up, because people are more likely to
eat ice cream and also more likely to go for a swim when it's hot out. When there is no working theory for how
things correlate, a good path forward is to investigate what other factors might contribute to the things that are
correlated. One way to begin that is to ask what characteristics might be shared.

When we see correlation, how do we know whether there is causation? Let’s say we have two data sets, A and
B, where we do have correlation (positive or negative). There are four possibilities for the correlation:

1. A is the cause of B.
2. B is the cause of A.
3. A and B are caused by some third thing, C, as with the ice cream sales and drownings.
4. The correlation is accidental (either nonsense or spurious).

In order to infer which situation we have when we encounter correlated data, we can rely on our general
background knowledge (i.e., things we know to be true about the world), our scientific knowledge, and possibly
further scientific testing.
For example, in the global warming case, there is no scientific theory that explains how rising global
temperatures could cause rising levels of , but there is a scientific theory that can explain the reverse causal
direction. This knowledge makes it plausible to infer that the rising levels cause the rising average global
temperatures.

 TERM TO KNOW

Accidental Correlation
When there is evident correlation between two variables but no explanation of how one can cause the
other or how both can be affected by the same third factor. There are two types of accidental
correlations: nonsense and spurious.

 SUMMARY

In this lesson, you learned more about causal theories. Evaluating causal theories requires rigorous
testing, which may mean controlling many variables over time to assure that you have a full
understanding of causes and effects. This may be complicated by concomitant variation, when it is
impossible to isolate or remove a potential cause or to measure effect as mere presence or absence,
usually because it is always present to some degree. It is also important to watch for accidental
correlation, when a seeming correlation is due to mutual effects of some third unknown variable
(spurious correlation), or simply a coincidence (nonsense correlation).

Source: THIS CONTENT HAS BEEN ADAPTED FROM Introduction to Logic and Critical Thinking.

 TERMS TO KNOW

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 18
 Accidental Correlation
When there is evident correlation between two variables but no explanation of how one can cause the
other or how both can be affected by the same third factor. There are two types of accidental correlations:
nonsense and spurious.

Concomitant Variation
The measure of two factors over time to see how one relates to the other.

Negative Correlation
When measuring two factors shows that one increases as the other decreases and vice versa.

Positive Correlation
When measuring two factors shows that both roughly increase and decrease at the same time.

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 19
 Inference to the Best Explanation
by Sophia

 WHAT'S COVERED

In this lesson, you will learn strategies for creating or evaluating hypothetical explanations for observed
phenomena. In particular, you will learn about:
1. Inference to the Best Explanation
2. Conditions of Good Explanations
2a. Explanatoriness
2b. Depth
2c. Explanatory Power
2d. Falsifiability
2e. Modesty
2f. Simplicity
2g. Conservativeness

1. Inference to the Best Explanation

You may recall that some deductive arguments provide explanations to help us understand why something
happened, not convince us that something happened. These arguments still have premises and a conclusion,
and follow the same logical rules, but it helps when reading arguments in natural language to know that they
are not always trying to persuade us that the conclusion is true. There is also a common kind of inductive
argument that is explanatory. Like all inductive arguments, explanations are not absolutely true or false. Just as
other inductive arguments might predict the most probable outcome, an explanatory inductive argument tells us
the most likely cause for something.

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 20
 Explanatory reasoning overlaps with causal reasoning. After all, both are about things happening that cause
other things to happen. However, explanatory reasoning is generally concerned with solving a single problem
or understanding how and why something happened once; we aren’t assuming that the same conditions will
cause the same event to happen again as we do with causal reasoning.

Imagine you find your car window broken and your phone missing. In trying to explain the event, you speculate
that someone broke your window to steal your phone. This explanation explains all the relevant facts (broken
window, missing phone) and furthermore seems likely. But there are other possibilities. Perhaps a stray baseball
broke the window, and then a dog jumped through the window and nabbed the phone. This probably doesn’t
feel as satisfying as an explanation, but why not?

Inference to the best explanation is a form of inductive argument where we consider a set of observed facts
and come up with the most likely explanation. That somebody broke your window to steal your phone is an
inference to the best explanation. Here is its form:

1. Observed facts: Your car window is broken and your phone is gone.
2. Explanation: The hypothesis that a thief broke the window and stole your phone provides a reasonable
explanation of the observed facts.
3. Comparison: No other hypothesis provides as reasonable an explanation.
4. Conclusion: Therefore, a thief broke your car window and stole your phone.

Notice that this is an inductive argument because the premises could all be true and yet the conclusion is false.
Just because something is reasonable doesn’t make it true. After all, sometimes things happen that defy
reason. So, perhaps the baseball-dog hypothesis was actually true. In that case, the premises of the argument
would still be true (after all, the thief hypothesis is still more reasonable than the baseball-dog hypothesis) and
yet the conclusion would be false. But inference to the best explanation arguments are not intended to be
deductive arguments, but inductive arguments, and as we learned earlier, inductive arguments don’t always
lead to true conclusions. When they are wrong, however, it should be because something extraordinary
happened.
However, what makes an explanation reasonable? What if your car was parked in front of a baseball field and
next to a dog park? What if the door was left unlocked and the phone was too old to be very attractive to
thieves? The notion of a “most reasonable explanation” is by nature subjective.

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 21
  TERM TO KNOW

Inference to Best Explanation

A form of inductive reasoning where we work from a set of observed facts to come up with the most
likely explanation for an event.

2. Conditions of Good Explanations

There are certain conditions that any good explanation must meet. The more of these conditions that are met,
the better the explanation.

2a. Explanatoriness
To meet the condition of explanatoriness, explanations must explain all the observed facts.

EXAMPLE The baseball explanation (by itself) does not explain the missing phone, so it is less attractive
as an explanation.

 TERM TO KNOW

Explanatoriness
Explanations must explain all the observed facts.

2b. Depth
To meet the condition of depth, explanations should not raise more questions than they answer.

EXAMPLE If the car is parked in an underground parking garage, the baseball explanation would lead
to additional questions like, who would be tossing a baseball around in an underground parking garage and
why?

 TERM TO KNOW

Depth
Explanations should not raise more questions than they answer.

2c. Explanatory Power

To meet the condition of explanatory power, explanations should apply in a range of similar contexts, not just
the current situation in which the explanation is being offered.

EXAMPLE If there were a rash of broken car windows and missing valuables, the best explanation
would not be that many baseballs were flying around and many unleashed dogs were leaping through
windows.

 TERM TO KNOW

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 22
 Explanatory Power
Explanations should apply in a range of similar contexts, not just the current situation in which the
explanation is being offered.

2d. Falsifiability
To meet the condition of falsifiability, explanations should be falsifiable, meaning there must be a way to test
the explanation for truth or falsehood. This does not mean that the theory can be disproven, just that it can be
truly tested. In this case, perhaps a theory is that the baseball came to life in the extreme heat of the car and
burst free of the windshield. However, imagine a further part of the theory is that such spontaneous genesis
never occurs when the baseball is being observed. So, if we attempt to replicate the event (to test the theory),
the test conditions themselves make it impossible. Non-falsifiable theories are not necessarily false, but they
give us nothing to do as critical thinkers. In the real world, many popular beliefs (such as psychic powers or
ghosts) are non-falsifiable, but good critical thinkers should be skeptical of any claim that cannot be tested.

 BIG IDEA

Recall the example from the previous tutorials where we described rigorous testing to find the causes for
our dog’s reactions. A key habit in critical thinking that applies to any form of inductive reasoning is being
able to come up with a way to test your theory.

 TERM TO KNOW

Falsifiability
The best explanations can be tested for truth or falsehood.

2e. Modesty
To meet the condition of modesty, explanations should not attempt to explain anything more than the observed
facts.

EXAMPLE It isn’t necessary to explain a rash of cat burglaries in the neighborhood where the phone
was (ostensibly) taken, only the single incident of the broken window and missing phone.

 TERM TO KNOW

Modesty
Explanations should not attempt to explain anything more than the observed facts.

2f. Simplicity
To meet the condition of simplicity, when all other things are equal, the simplest explanation is the best—the
one with the fewest entities and events. This is often referred to as Occam’s Razor (in philosophy, a razor is a
rule of thumb that allows a thinker to eliminate, or shave off, unlikely explanations for a phenomenon). For
example, the baseball and dog explanation involves two entities (one baseball, one dog) and two events (the
window breaking, the dog getting into the car), where the thief hypothesis requires only one entity and one
event.

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 23
  TERMS TO KNOW

Simplicity
The better explanation is the one with the fewest entities and events.

Occam’s Razor
All other things being equal, the simplest explanation is best. The simplest explanation is the one
requiring the fewest entities and events.

2g. Conservativeness
To meet the condition of conservativeness, explanations that force us to give up fewer well-established beliefs
are better than explanations that force us to give up more of them. For example, say the person with the stolen
phone is visiting a friend, and somebody suggests the friend actually stole the phone. This requires the victim of
the crime to question a valued friendship, and what he knows about the person.

It’s crucial to see how each of these are used not to come to the only possible true conclusion, but to the most
likely one. For example, there are many cases, especially in science, where the right explanation is not so
simple. Indeed, science must often posit new, sometimes purely theoretical, entities.

EXAMPLE As early as the 1600s, some scientists speculated that there were “little animals” or “worms”
that caused illness and spread from person to person, even across great distances. This idea introduced a
new entity, a microorganism, that would not be directly observed for another couple of centuries and at the
time could not be rigorously tested. However, these scientists were seeking ways to explain things that
could not be adequately explained with simpler premises. Moreover, they actively pursued ways to find out
more and test germ theory with the materials they had.

 TERM TO KNOW

Conservativeness
The better explanation is the one that forces us to give up fewer established beliefs.

 SUMMARY

In this lesson, you learned about inference to the best explanation, the practice of finding the most
reasonable or plausible explanation for events. These explanations don’t claim to be the only possible
answer, but the most probable or likely answer. The conditions of good explanations include seven
good habits for critical thinking, including fully explaining the phenomenon (explanatoriness), the
explanation itself not raising more questions than it answers (depth), and being generalizable to new
situations (explanatory power). A good explanatory theory should be able to be tested (falsifiability),
and not be too broad (modesty). When presented with more than one possible explanation, we opt for
the one that involves the fewest entities and events (simplicity) and challenges the fewest well-
established beliefs (conservativeness).

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 24
 Source: THIS CONTENT HAS BEEN ADAPTED FROM Introduction to Logic and Critical Thinking.

 TERMS TO KNOW

Conservativeness
The better explanation is the one that forces us to give up fewer established beliefs.

Depth
Explanations should not raise more questions than they answer.

Explanatoriness
Explanations must explain all the observed facts.

Explanatory Power
Explanations should apply in a range of similar contexts, not just the current situation in which the
explanation is being offered.

Falsifiability
The best explanations can be tested for truth or falsehood.

Inference to Best Explanation

A form of inductive reasoning where we work from a set of observed facts to come up with the most likely
explanation for an event.

Modesty
Explanations should not attempt to explain anything more than the observed facts.

Occam’s Razor
All other things being equal, the simplest explanation is best. The simplest explanation is the one
requiring the fewest entities and events.

Simplicity
The better explanation is the one with the fewest entities and events.

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 25
 Analogical Reasoning
by Sophia

 WHAT'S COVERED

In this lesson, you will learn about analogical reasoning, the form of inductive argument that looks for
comparisons between well-known things and a lesser-known thing to aid in understanding. Specifically,
this lesson will cover:
1. Argument From Analogy
2. False Analogies
3. Disanalogies

1. Argument From Analogy

Another kind of common inductive argument is an argument from analogy. In an argument from analogy, we
note that since one thing is like another thing, it must have some of the same qualities.

For example, suppose a person is buying a new car. They have owned a Subaru in the past and found it to be
reliable, so they assume a new Subaru will also be reliable. We can formalize this analogical argument as
follows:

1. The old Subaru was reliable.

2. Therefore, the new Subaru will also be reliable.

Understanding this argument as a statistical generalization is problematic because of the small sample size: the
single piece of data (previously owning one reliable Subaru) could have been a fluke. Knowing the reliability of
a single instance makes a generalization about the entire class of Subarus, i.e., “Subarus are reliable cars,” a
weak one.
However, the inductive argument can be strengthened. Suppose the owner instead has previously owned
multiple Subarus. They have observed that the manufacturer puts an emphasis on reliability over style and
knows that the new car has the same engine and transmission as the earlier vehicles. The person may also
know that there’s been no significant changes in the company since the last car was bought.

All this information adds premises that strengthen the analogy between the old Subaru and the new one. Let’s
formalize this update:

1. My old Subarus were all reliable.

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 26
 2. Subaru values reliability over style.
3. The new Subaru and the old one have the same engine and transmission.
4. There have been no significant changes in the company between the newest and most recent previous car
models
5. Therefore, the new Subaru will be reliable.

Notice all these other analogies between the old and new Subarus, and the analogies between how the
company was previously and is now, give credibility or strengthen the analogy about the reliability of the new
Subaru. You might even analogize this situation to a person returning to a restaurant that has always had good
food and service.
In short, this is now not a hasty generalization from a random sample of cars made by Subaru, but a strong
analogical argument between like things. The old cars and the new car are made by the same company, which
is known to be reliable. The new car has some of the same parts and the same assembly. In short, an inductive
argument by analogy can be strong even if it is based on a small sample because there are deeper and more
specific relations between the two things.

This is a key condition for any good argument from analogy: the similar characteristics between the two things
cited in the premises must be relevant to the characteristic cited in the conclusion.

 BIG IDEA

The keys to an analogy being effective are:

1. The number of characteristics the object and the analog have in common.
2. The depth of the relationship between the object and the analog.
3. The relevance of the analog to the current situation.

For a recent example, in late 2019, the threat of COVID-19 was made analogous to the threat of SARS or avian
flu, because those infectious disease outbreaks shared some characteristics with COVID. Yet, by spring 2020, it
was clear that the most relevant analogy was actually the 1918 influenza epidemic. In fact, much of our everyday
decision making is driven by analogies: we make a set of assumptions about something based on the
characteristics it shares with things familiar from our past experiences.

 THINK ABOUT IT

Have you ever told someone, or been told by someone, that a decision you’re about to make is “just like
last time”? It might be getting overly excited for long-shot job opportunities, dating people who aren’t good
matches, or other behaviors where an analogy can be drawn between past outcomes and the present that
might lead you to a conclusion about whether to repeat the behavior.
We’ve been discussing analogy as a kind of inductive argument. Analogical thinking is also a key aspect of the
psychology of decision making. It helps us to make predictions about future experiences by drawing analogies
to past experiences. Analogical thinking is heuristic, meaning it is a shortcut that often (though not always) leads
to reliable results. Good critical thinking involves questioning assumptions, and assuming that characteristics
are shared (and especially that the analogy is fitting) is indeed an assumption. As we have reiterated throughout
this course, we should thus be ready to question our own assumptions. So, we must question the analogies we
and others give for accuracy and good faith.

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 27
  TERM TO KNOW

Argument From Analogy

An inductive argument where one or more perceived similarities between two things are presumed to
indicate deeper similarities.

2. False Analogies
Analogies can be powerful and effective, but fallacies may be disguised as false analogies, which attempt to
draw on the power of a past event or popular creative work when the comparison is weak or inapt.

Drawing an analogy can be a rhetorical tactic for undermining an argument. Whether the analogy drawn is true
or false depends on the relevance and strength of the (purportedly) shared similarities. Often analogies to
traumatic historical events play this role because appeal to them is highly contentious and inflammatory.

“This is the new McCarthyism”: The practice by the infamous Senator Joe McCarthy of hunting and
prosecuting (“blacklisting”) purported communists in the 1950s, effectively destroying their careers and
lives.
“This is a witch hunt!”: The medieval practice of putting people on trial for witchcraft, and in the U.S.,
generally understood to be invoking the Salem Witch Trials of 1692.
“This is like Jim Crow all over again”: The legalized system of discrimination against Black Americans in the
American South from the 1870s through the 1960s.

However, whether any of these count as a false analogy will depend on their relationship to what they’re being
compared to. Are there any highly relevant dissimilarities? How relevant and strong are the similarities?

 DID YOU KNOW

In 1953, the playwright Arthur Miller wrote The Crucible, a dramatization of the Salem Witch Trials, that was
intended as a statement by analogy about the efforts of the McCarthy hearings to weed out communists in
American life at the time. Although the play is set in the 17th century and makes no reference to current
events, the analogy was strong enough—through the similar tone and situations of the trials as shown in the
play and in the news—that audiences understood what the play was really "about." The analogy was so
strong that Miller was even prosecuted by McCarthy in relation to the play and his career in the
entertainment industry. Miller’s refusal to name suspected communists garnered him a conviction for
contempt of court from the House Committee on Un-American Activities in 1956. Miller was also the
husband of the blonde bombshell actress Marilyn Monroe.
Analogies to infamous fictional characters, places, and events can also play the same role as analogies to the
historical events above, because of the purported villainy of these characters. And as before, whether the
analogy is false depends on the relevance and similarities of the comparison.

“This new government program smacks of Big Brother” (drawing an analogy to the dystopian novel 1984 by
George Orwell).

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 28
 “That politician reminds me of Darth Vader/Voldemort/Lex Luther” (drawing analogies to iconic villains from
movies, books, and comics).

EXAMPLE In 2022, Arizona senator Andy Biggs used the word “Orwellian” to describe an investigation
of white supremacy in the police and military despite the fact that unchecked police power is a much more
recognizable characteristic of the novel than police being investigated. Such a strong dissimilarity makes
this indicative of a false analogy.
False analogies are not always negative. Favorable analogies can be false too:

Sure, you laugh at my time machine. Well, they laughed at Galileo, they laughed at Isaac Newton, and they
laughed at Thomas Edison. We’ll see who laughs last!

Again, even if we ignore any historical inaccuracies here, the aptness of the analogy comes from finding more
specifics than someone being mocked. The obvious dissimilarity between the evidence and success of the
speaker and these famous scientists should be a red flag.
The strength of an analogical argument relies on how many characteristics the two things have in common and
how deep those differences are. We can also consider the scale of the two things being compared, such as a
trivial event being compared to a major historical event. We can also ask if the analogy is being made in good
faith. Are we, or the person making the argument, convinced that these two events are similar, or are they
raising the specter of evil to demonize their opposition? The answer to all of these questions matter for whether
the analogy is false or good, a fallacy, or a strong inductive argument.

 TERM TO KNOW

False Analogy
An analogy that draws a broad comparison between two things that, upon scrutiny, share few
characteristics, do not have any deep resemblance, and any shared characteristic is not relevant to the
situation.

3. Disanalogies
An analogy can be hard to refute directly. But we can point out disanalogies. A disanalogy is a crucial difference
between the object under discussion and the analog.

In 2021, with the rise of the vaccine mandate during the COVID-19 epidemic, some people who refused to get
the vaccine briefly wore yellow stars as a kind of analogy to Jews in Nazi Germany. They were stating that they,
like the Jews in Nazi Germany, were being persecuted for their beliefs. The disanalogy is that Jews in Nazi
Germany were forced to wear the insignia, while anti-vaxxers choose to do so. This is a crucial disanalogy. How
can choosing to do something be “exactly like” being forced to do it?

No two things will have absolutely everything in common, so disanalogies must be fundamental to the
comparison to be effective.

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 29
  TRY IT

Supporters of mask mandates during COVID also used historical analogies, comparing the mandate to
mandatory seatbelt laws and smoking bans. They were also compared to store policies like “no shirt, no
shoes, no service.”

Are these fair analogies or false analogies? Consider the common characteristics and the
+
differences.

These are all more defensible analogies because they are similar in scope to mask mandates and have
common characteristics of imposing requirements on people for going out in public. However, you may
have seen differences. Though this a difference between mask mandates and seatbelt laws in terms of
legal consequences for noncompliance, this is not a crucial dissimilarity because the comparison is
about basic restrictions on personal autonomy for public safety. Or you might see that people can go
through a meal without smoking, but they can’t go through a meal with their mask on, and removing
the mask while they are eating compromises the policy.

 TERM TO KNOW

Disanalogy
A crucial difference between two things being compared, used to refute a false analogy.

 SUMMARY

In this lesson, you learned about analogical reasoning. An argument from analogy can be a strong
inductive argument when constructed well. Arguments from analogy are also used heuristically in
making decisions based on the visible characteristics of something and your past experience. We also
saw false analogies, which are inapt or inaccurate comparisons, sometimes given in bad faith. We can
use disanalogies, the crucial discrepancies and inconsistencies between two things, to expose false
analogies.

Source: THIS CONTENT HAS BEEN ADAPTED FROM Introduction to Logic and Critical Thinking.

 TERMS TO KNOW

Argument From Analogy

An inductive argument where one or more perceived similarities between two things are presumed to
indicate deeper similarities.

Disanalogy
A crucial difference between two things being compared, used to refute a false analogy.

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 30
 False Analogy
An analogy that draws a broad comparison between two things that, upon scrutiny, share few
characteristics, do not have any deep resemblance, and any shared characteristic is not relevant to the
situation.

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 31
 Statistical Generalizations
by Sophia

 WHAT'S COVERED

In this lesson, you will learn how to evaluate statistical generalizations that are premises in inductive
arguments. In specific, you will learn how to consider and evaluate:
1. Statistical Generalizations
2. Evaluating Statistical Generalizations
2a. Sampling
2b. Recency and Relevance

1. Statistical Generalizations
Statistics is a branch of math concerned with uncovering patterns from observed data, and reasoning about the
possible causes of those patterns. You may notice that all our examples of inductive arguments include
statements like these:

It’s not likely that a Major League Baseball team can win a 7-game series after losing the first three games.
The odds that Tiger Woods will score under par is high.
Most healthy, normally functioning birds can fly.

If we look up past baseball playoffs, and counted how many teams have won a series after trailing by three
games, we might find that the first statement was true. If we track Tiger Woods’ golf game, we can calculate the
average par for each hole and make a conclusion like the second. Finally, if we observe birds in the wild, we
might make the generalization in statement 3. Note, in each case, there can be exceptions: like the Red Sox,
Tiger Woods in the 2022 Masters (he shot way over par); ostriches and emus are normally functioning birds that
cannot fly.
Each of these is a statistical generalization, a statement about how probable or likely an event is based on
empirical evidence. Empirical evidence is information that can be observed (or detected through other senses)
and measured. When we have collected data, we can use mathematical computations (based on probability
theory) to make generalizations about the patterns found within that data.

 TERMS TO KNOW

Statistical Generalization
A statement about how probable or likely something is based on empirical evidence.

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 32
 Empirical Evidence
Information that can be observed (or otherwise detected by the senses).

2. Evaluating Statistical Generalizations

Consider the following inductive argument, whose premise is a statistical generalization:

1. Seventy percent of likely voters say they will vote for Stevens.
2. Therefore, Stevens will probably win the election

This might be shared in good faith, but there are many ways political polling can fail. For example, perhaps the
pollster only called a hundred people in a population of millions. Or perhaps some aspect of the polling itself—
how they contact people, who they contact, or how they ask the question—skews the result.

EXAMPLE In 1948, newspapers ran stories that Thomas Dewey had won the presidential election,
based on polling showing that he was well ahead of Harry Truman. In fact, Truman won the election. The
pollsters made a grave error; they conducted the poll completely by telephone at a time when many poorer
households did not have a phone. This led to overcounting Republican voters and under-counting
Democratic voters.
But while political polling is the most prone to error, other statistical generalizations can also misrepresent facts
or mislead people. We can assess whether or not a statistical generalization is good support for an argument by
considering whether the statistical generalization meets certain conditions. Some of the considerations are
below.

2a. Sampling
Many statistical errors result from poor sampling. A sample is simply a portion of a population from which a
statistical generalization is made about the whole population. A population is usually a group of people but may
also refer to a collection of objects or events. For example, transportation departments often do studies to
count the numbers of different kinds of vehicle that use a certain road or highway; these are statistical samples
of the population of vehicles. Other statistics might measure events, such as in sports, where they track batting
averages, football passes, or shots on goal. These are populations of in-game events. When it’s impossible to
count every single member of the population, sampling is used, a method for measuring a slice of the whole
population.

There are two conditions that a sample must meet in order to provide useful information about the whole:

1. Adequate sample size: the sample size must be large enough to support the generalization.
2. Non-biased sample: the sample must be representative of the population and not biased.

First, a sample size must be large enough to draw a generalization. For example, say that someone is curious
how many people have cats versus dogs, and observes shoppers at a pet supplies store to find out. They see a
dozen people. Of these, eight buy cat food, three buy dog food, and one buys bird food. Based on this sample,

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 33
 they conclude that far more people have cats than other pets. However, twenty minutes later, the observer
might have seen more people buying dog food—the sample might have changed. It is hard to draw meaningful
conclusions from small sample sizes.
Fortunately, there is an easy mathematical rule to apply here: the sample size should be at least 10% of the
population, but only up to a sample size of 1,000. For example, if you want to sample from an organization of
3,000 people, you need to survey 300 people. However, if you want to sample from a state with three million
inhabitants (say, Arkansas), you only need to survey 1000 people. This may seem astonishing, but over a large
population, a relatively small sample can be used to make statistical inferences, provided it is random enough
(sampling everyone). This is just a rule of thumb for determining sample size.

 TERMS TO KNOW

Sample/Sampling
The portion of the population that is being directly observed or assessed to make a generalization, or
the process by which that sample is obtained.

Population (in Statistics)

The group of things or events that are described by a generalization.

2b. Recency and Relevance

Second, a good statistical generalization is also one in which the sample is representative of the population,
meaning that the sample has the same characteristics as the whole (it represents the characteristics of the
population). For example, the traffic study may not get a representative sample if they watch a road from 10 a.m.
to 2 p.m. because there may be different traffic patterns during those hours. Perhaps there are more trucks as
service and delivery vehicles make their rounds, and fewer passenger cars because the study is conducted
only during the working hours of the day. The sample taken would be different from the characteristics of
vehicles that travel on the road throughout the full day. This is called a biased sample. It doesn’t mean that
there is necessarily inherent bias on the part of the person drawing the conclusion, but because of how the
data was collected, some part of the population is over-counted, and another is under-counted.

 THINK ABOUT IT

Some municipalities will conduct annual bike counts to determine where and when people ride bicycles. A
common approach for a bike count would be to have volunteers stand at a high-traffic street corner in
shifts, counting all the bikes that go by in each direction.

How would you design a bike count to ensure that the count is representative of cycling for the entire
population? What hours and days of the week would you want to observe? Would you station volunteers in
one busy location, or several strategic spots? How might weather impact the bias of the bike count's
sampling?
Making a broad conclusion about a sample that is not large enough or representative is an informal fallacy
called hasty generalization. Hasty generalization fallacies are very common in everyday discourse, as when a
person gives just one example of a phenomenon occurring and treats it as sufficient evidence for a
generalization.

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 34
 EXAMPLE A person might say they will never buy another Toyota because their uncle had one and it
was a lemon. While no manufacturer is perfect, the truth is that Toyota is one of the most reliable car
manufacturers on the market, so this person is making a hasty generalization with a sample size of one.
Importantly, they are also relying on hearsay without the deep knowledge to make an analogy.
You may see that a hasty generalization leads to illusory correlation bias, which we learned about in an earlier
unit. Both are due to a tendency to overvalue anecdotal evidence because your personal experiences (or the
firsthand experiences of people you know) tend to have more psychological weight than abstract figures.

There are several ways that the quality of sampling can impact the reliability of a statistical generalization. If the
sampling is out of date, then it is less reliable than if it includes recently sampled data. Consider the following
generalization:

Women make 61 cents for every dollar made by men doing the same job.

You might suspect the data is old and note that observable changes have recently been made in gender equity.
In fact, this reports a statistic that was true in 1960. As of 2020, women made 83 cents for every dollar made by
men doing the same job. But sampling in this case is extra crucial. Not all women’s material conditions are the
same. Depending on what women’s salaries you pull along racial, economic, and class lines, the resulting
generalizations can vary wildly. Women of color specifically are at an even greater disadvantage—they earn
only 63 cents for every dollar a white man is paid for the same job.
For any statistical generalization, we must ask: Who compiled the statistic and how? When and where was the
data collected? Perhaps a statistic is legitimate but out of date. Or perhaps it was done recently, but in a
different cultural context. Whether these “count” depends on the goals of the generalization. All these can
affect the quality of a statistical generalization.

There is no pat answer here, except to ask good questions. However, this doesn’t mean trying to find a way to
dismiss any research that challenges your opinion!

 TERMS TO KNOW

Representative (Sample)
The quality of a sample being a good measure of the population, such as counting across demographic
lines in a poll.

Biased (Sample)
The quality of a sample not being a good/representative measure of the population, by under-counting
or over-counting some members. This may be intentional or unintentional.

Hasty Generalization
Making a statistical generalization on a sample that is either inadequate in size or not representative of
the whole.

Anecdotal Evidence
Making statistical generalizations based on first- or second-hand experience.

 SUMMARY

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 35
 In this lesson, you learned about statistical generalizations, or the science of drawing conclusions from
partial information. While statistical generalizations can lead to accurate predictions, it’s important to
know how to evaluate statistical generalizations and know when to trust them. The reliability of
statistical generalization largely relies on sampling; it is critical that the sample size is large enough and
representative of the whole. Other criteria to consider are recency and relevance, as well as knowing
which are gathered in good faith and not skewed on purpose.

Source: THIS CONTENT HAS BEEN ADAPTED FROM Introduction to Logic and Critical Thinking.

 TERMS TO KNOW

Anecdotal Evidence
Making statistical generalizations based on first- or second-hand experience.

Biased (Sample)
The quality of a sample not being a good/representative measure of the population, by under-counting or
over-counting some members. This may be intentional or unintentional.

Empirical Evidence
Information that can be observed (or otherwise detected by the senses).

Hasty Generalization
Making a statistical generalization on a sample that is either inadequate in size or not representative of
the whole.

Population (in Statistics)

The group of things or events that are described by a generalization.

Representative (Sample)
The quality of a sample being a good measure of the population, such as counting across demographic
lines in a poll.

Sample/Sampling
The portion of the population that is being directly observed or assessed to make a generalization, or the
process by which that sample is obtained.

Statistical Generalization
A statement about how probable or likely something is based on empirical evidence.

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 36
 Probability (Part 1)
by Sophia

 WHAT'S COVERED

In this lesson, you will begin to learn about probability, a branch of mathematics that involves
measuring chances of future events. Specifically, this lesson will cover:
1. Calculating Probability
2. Probability of Conjunctive Statements
3. Probability of Disjunctive Statements

1. Calculating Probability
By now you understand that an inductive argument involves statements about what is likely or probable. How
do we determine whether something is likely or probable?

To answer this, we need to quantify the things in question; that is, we need to be able to count or measure the
events communicated in our statements. Consider a standard deck of 52 cards, and the statement below:

You’re not likely to draw a Queen.

It’s probably going to rain tomorrow.

These are probabilistic statements about the likelihood of a future event occurring. However, what does it mean
for an event to be “likely” or “probable”? These terms don’t actually give us a concrete measure on their own.
Actually, it’s straightforward to measure the probability of drawing a Queen. All we need to know are the
parameters of the possibility space: how many cards are there total and how many of those are Queens.

Probability is the measure of how likely it is that something is (or will be) true, or how likely it is for a given event
to occur. A probability, p, will always be a number between 0 and 1. If p=0, the event will never occur or is
impossible, while if p=1, the event will always occur. We frequently express probabilities as a percentage, a rate
of occurrence out of 100 times (“per” means “for” and “cent” means 100 in Latin). In percentage, 0% means the
same as a probability p=0, and 100% means the same as probability p=1.

The probability of a single event is calculated by counting the total number of possible events (possibilities) and
the total number of those events that match the event we want to measure (outcomes). We write this formula as
follows: P(Event) = P(Outcomes)/P(Possibilities). We derive the probability of our event by dividing the outcomes

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 37
 by the possibilities. Because the possibilities must be equal to or greater than the number of outcomes, and
because they must be positive numbers, this calculation will always give us a number between 0 and 1.

 FORMULA TO KNOW

Let’s return to a deck of cards. Imagine your friend picks one at random. How would we calculate the probability
that they have picked a Queen? There are 52 cards in the deck, so we have 52 possibilities. There are 4
Queens in a deck, so we have 4 possible chances to get a Queen. So, the probability of picking a Queen is 4
out of 52, 4/52, or in decimal form as the probability, p=.077. Note that we round our decimals to two places.

To express these chances as a percentage, we simply multiply the decimal by 100. Your friend has a 7.7%
chance of picking a Queen. In comparison, the chances of picking any face card (King, Queen, or Jack) are
much higher. There are three face cards in each suit and four suits, which means there are 3 x 4, or 12 face
cards total. So, the probability of a face card is 12/52 = .23, or 23%.

The odds of an event are represented by the ratio of the probability of success (the probability of the event
occurring) to the probability of failures (the probability of the event not occurring). We know already that the
probability of success for choosing a Queen is 4/52. What is the probability of failure? Remember that if we
added together the probability of each outcome (or summed them), the total would be 1. Given this, the
probability of failure is the inverse of the probability of success. If there are 4/52 ways to succeed, it means that
all other times, we would fail. So, “all other times” is calculated by subtracting the successes from the total: 52-4
= 48, or by subtracting the probability of the successes from 1. Since the odds are the ratio between the two, the
odds of choosing a Queen are 4/52 divided by 48/52, or 4/48.

 TERMS TO KNOW

Quantifiable
The ability to be counted or measured.

Probability
How likely a particular event will occur.

Percentage
Literally “in 100,” all probabilities can be expressed as a percentage.

Possibilities
The total possible events, or outcomes.

Outcome
The possible event whose probability we are trying to determine.

Odds
The ratio of the probability of success to the probability of failures. You can calculate one value from the
other by subtracting the probability you know already from 1. Since the possible outcome probabilities
must sum to 1, the outcome you don’t already know has to be the remaining probability.

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 38
 2. Probability of Conjunctive Statements
We now know how to calculate the odds of one thing happening if we know the number of total possibilities
and the number of outcomes we want to measure. But what about the probability of two things happening? Or
of one thing happening at least once across several tries? Or of something not happening?

Let’s start with multiple things all occurring. First, it’s important to understand that when we calculate the
probability of multiple events occurring, we are essentially computing the probability of a conjunctive statement:
that event A occurs and that event B occurs and that event C occurs, and so on.

The probability of the conjunction of two events is less likely than the probability of each single event. This
should make intuitive sense. Recall the truth table for conjunction: a conjunction is true in only one scenario out
of the four; assuming that our premises are either certainly true (p=1) or certainly false (p=0), the probability that
the conjunction is true is 1/4, or .25. This is why in probability theory, to determine the probability of a
conjunction, we multiply the probabilities of the conjuncts. When we multiply probabilities, our resulting
probability is smaller (technically, less than or equal to, but let’s not get caught up on that).

It helps to know whether the events are independent (meaning that neither event affects the other) or
dependent (meaning that the outcome of one event affects the likelihood of the other occurring).

For example, if we’re aiming to draw a Queen from the deck, and we draw twice, did we put the first card back
or not? If we put it back, it is called sampling with replacement; if we did not, we have sampled without
replacement. If we have sampled with replacement, then the two card-drawing events are independent
because the first draw doesn’t affect the probability of any outcome in the second draw.

If we have drawn without replacement, i.e., we didn’t return the card back to the deck, then that first draw has
now affected the possible outcome of the second draw. For example, if you had drawn a Queen that first time,
then we’d be one Queen down on the second draw with only 51 cards left. So, our previous 4/52 chances are
now 3/51 on the second draw (note that our chances of drawing a Queen substantially decreases). However, if
we didn’t draw a Queen on the first draw, then we could still have 4 possible successes, but out of 51, not 52,
possible outcomes on the second draw. The chances of the second event are dependent on the first if we’re
drawing without replacement.

To calculate the probability of multiple events, we just multiply the probability of each separate event. This is
called the General Conjunction Rule of probability: P(A and B) = P(A) x P(B).

If the events are independent, then the denominators will be the same for each event; if the events are
dependent, then the denominators will be different. Let’s start with independent events, where we return the
first card to the deck (sampling with replacement). The probability of drawing a Queen on the first try (as we saw
above) is .077, or 7.7%. The probability of drawing a second Queen are also 7.7% because the deck of cards

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 39
 hasn’t changed from the first draw (i.e., there are still 52 cards total, and 4 Queens, 4 Kings, etc.). So, the
probability of getting two Queens in a row is .077 x .077 for .006. This is 0.6% (or one in 167 tries).

 WATCH

View a demonstration of this example.

 HINT

If you divide 1 by your final answer, the number you get can be plugged in to the phrase “one in ____,”
which is often easier to communicate. Remember to use the decimal number in this calculation and not the
percentage!
But if we don’t replace the first card in the deck, the second draw will be dependent on the first, specifically on
which card (i.e., outcome) has been removed. There are 51 cards left, but how many Queens are left? That
depends on whether we drew one the first time. There are either three or four Queens left. How do we know
which possibility to use in the second calculation?

If we’re only thinking about the probability that both draws are Queens, we don’t care about outcomes where
the first draw was not a Queen. Rather, we only count cases where a Queen was drawn the first time because
the probability that we’re calculating requires two successful Queen draws. The probability of drawing a second
Queen are then 3/51, which is .059.

Now that we have the probability of drawing two Queens (without replacement), we can calculate the
probability of the events occurring together. Let’s write out each step clearly.

1. P(Queen1 and Queen2 without replacement) = P(Queen1) x P(Queen2 without replacement)

2. P(Queen1) = 4/52, P(Queen2 without replacement) = 3/51
3. P(Queen1 and Queen2 without replacement) = 4/52 x 3/51
4. 4/52 x 3/51 = .077 x .059 = .004

So, the probability of two Queen draws (without replacement) is .004 or 0.4% (or one in 250 tries). While neither
event is particularly probable, notice that sampling with replacement actually gives you a higher probability of
drawing two Queens in a row because the probability of the second event is higher with replacement than
without.
You can use the same formula for more than two events, calculating the probability of each event and
multiplying them all together. Consider whether the events are dependent or not, so you can calculate the
probability of each event properly.

 TRY IT

Use a calculator to find the answer and express it as a percentage.

What is the probability of drawing four Queens in a row if you sample without replacement? +

The odds of drawing four Queens in a row is 0.000004, or 0.0004%. These numbers are so small they
are hard to communicate, but you can then divide one by 0.000004 and say the probability is one in

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 40
 270,725, which is a bit easier to comprehend!

Let’s look at another example. Suppose we read a story about a person who has been struck by lightning twice
in their lifetime. The probability of being struck by lightning in one’s lifetime (presuming one is not especially at
risk) are about one in a million. Since being struck by lightning the first time and the second are independent
events (what would it mean for them to be dependent?), we can calculate the probability of being struck by
lightning twice as 1/1,000,000 x 1/1,000,000. We end up with 0.000001 x 0.000001, which is 0.000000000001,
or 0.0000000001%. This is one in a trillion. Yet it happens with more frequency than that because some people
have high-risk jobs that increase their chances of being struck by lightning.

 TRY IT

What is the probability of rolling a 12 on two normal six-sided dice? +

The only way to roll a 12 on two normal dice is to roll a six on each one. The probability of rolling a six
on a single die is 1 in 6, so the probability of rolling two sixes is 1/6 x 1/6, or 1/36. Expressed as a
decimal, this is 0.028 (rounded), or 2.8%.

 TERMS TO KNOW

Independent
In probability, two or more events having no effect on one another.

Dependent
In probability, when the outcome of at least one event has an effect on at least one other event.

Sampling With Replacement

In probability theory, resetting the conditions after each outcome, such as replacing each card drawn to
the deck or each numbered ball to a bin.

Sampling Without Replacement

In probability theory, having multiple outcomes without resetting the conditions, such as drawing
successive cards from a deck or numbered balls from a bin.

General Conjunction Rule

P(A and B) P(A) P(B)

3. Probability of Disjunctive Statements

How do we calculate the probability of one event or another? For example, drawing either a Queen or a King
from the deck on one draw? What about drawing a Queen or a heart?

Recall the truth table for a disjunction: a disjunction can be true inclusively (p v q) 3/4 times (p=.75) or exclusively
((p v q) ∧ ¬(p ∧ q)) 2/4 times (p=.5).

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 41
 “Exclusive or” in probability is called mutually exclusivity. A card can’t be both a King and a Queen; these
outcomes are mutually exclusive. To calculate “exclusive-or” disjunctive outcomes, we simply add together the
probability of the two possible outcomes. This is the General (Exclusive) Disjunction Rule of probability: P(A or
B) = P(A) + P(B).

 FORMULA TO KNOW

As we already determined, the probability of drawing a Queen is 7.7%. The probability of drawing a King is the
same, 7.7%. So, we can simply add these and see the probability of getting one or the other is 15.4%.

The “inclusive or” meaning rules in a third option that the “exclusive or” meaning doesn’t have. This actually
makes the “inclusive or” less probable than the “exclusive or,” as we saw with conjunction where the probability
of the conjunction is less or equal to the probability of either conjunct. So, to capture mutual exclusivity, we
don’t subtract the conjunction; but to capture the “inclusive or,” we do subtract the conjunction (the product of
the probabilities) since this outcome is less likely. The General (Inclusive) Disjunction Rule of probability is
expressed as P(A or B) = (P(A) + P(B)) – (P(A) x P(B)).

What is the probability of drawing a Queen or any card from the suit of hearts? This example is not mutually
exclusive because we might draw the Queen of Hearts. In this case, we subtract any outcomes which meet both
criteria, so they are not counted twice. (We will now use the variable E to represent “Event.”)

 FORMULA TO KNOW

We know the probability of getting a Queen is 4/52 (.077), and the probability of getting a heart card is 13/52
(.25). The probability of getting the Queen of Hearts is 1/52 (.019). The probability of getting either a Queen or a
heart is thus .308, or 30.8%.

What if we wanted to know the probability of drawing a Queen on multiple tries? For example, consider the card
game called “Three Card Monte” or “Find the Lady.” The dealer has three cards, and exactly one is a Queen.
The dealer shuffles the three cards, and the player tries to pick the Queen. (In reality, this is often a con, but
we’ll presume for our purposes that the game is fair!)

If it’s a fair game, the probability of winning is one in three. But say you are skeptical, and the dealer wants to
entice you to play. “I’ll give you two tries,” they say. So now you have a one in three chance of winning, and only
have to win once on two tries. How would you calculate the probability under these circumstances?

We will let you ponder this problem and come back to it in a bit.

 TERMS TO KNOW

Mutually Exclusive
When measuring the odds of one of multiple events occurring, this refers to events that cannot happen
simultaneously.

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 42
 General (Exclusive) Disjunction Rule
P(A or B) P(A) P(B)

General (Inclusive) Disjunction Rule

P(A or B) (P(A) + P(B)) - (P(A) x P(B)).

 SUMMARY

In this lesson, you began to learn about the methodology for measuring the likelihood of events
occurring. We started with the basic formula for calculating probability of a single event. To measure
this, we must know the total number of possible outcomes and the total number of outcomes that we
are trying to measure. We also discussed how to measure the probability of the conjunction of events,
by multiplying the probability of each individual event. To measure the probability of the disjunction of
events, you can simply sum the probabilities of each event if they’re mutually exclusive, or sum them
and then subtract the probability of the conjunction if they are not mutually exclusive.

Source: THIS CONTENT HAS BEEN ADAPTED FROM Introduction to Logic and Critical Thinking.

 TERMS TO KNOW

Dependent
In probability, when the outcome of at least one event has an effect on at least one other event.

General (Exclusive) Disjunction Rule

P(A or B) = P(A) + P(B)

General (Inclusive) Disjunction Rule

P(A or B) = (P(A) + P(B)) - (P(A) x P(B)).

General Conjunction Rule

P(A and B) = P(A) x P(B)

Independent
In probability, two or more events having no effect on one another.

Mutually Exclusive
When measuring the odds of one of multiple events occurring, this refers to events that cannot happen
simultaneously.

Odds
The ratio of the probability of success to the probability of failures. You can calculate one value from the
other by subtracting the probability you know already from 1. Since the possible outcome probabilities
must sum to 1, the outcome you don’t already know has to be the remaining probability.

Outcome

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 43
 The possible event whose probability we are trying to determine.

Percentage
Literally “in 100,” all probabilities can be expressed as a percentage.

Possibilities
The total possible events, or outcomes.

Probability
How likely a particular event will occur.

Quantifiable
The ability to be counted or measured.

Sampling With Replacement

In probability theory, resetting the conditions after each outcome, such as replacing each card drawn to
the deck or each numbered ball to a bin.

Sampling Without Replacement

In probability theory, having multiple outcomes without resetting the conditions, such as drawing
successive cards from a deck or numbered balls from a bin.

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 44
 Probability (Part 2)
by Sophia

 WHAT'S COVERED

In this lesson, you will continue to learn and use formulas for calculating probability. Specifically, this
lesson will cover:
1. The Probability of Something Not Happening
2. One Event Happening on Multiple Tries
3. The Probability of Winning the Lottery
4. The Birthday Problem

1. The Probability of Something Not Happening

We have learned how to calculate the probability of something happening, even the probability of two things
happening. But how do you calculate the probability that something will not happen?

This brings up a simple property of probability. The probability of something happening is expressed as a
number between 0 and 1. The range between 0 and 1 represents all possibilities. Thus, the probability of an
event not happening is 1 minus the probability that it will happen. This is also called the inverse.

 FORMULA TO KNOW

For example, suppose you are playing a board game and want to determine the probability that you will not roll
a 12. We saw previously that rolling a 12 on two normal dice is 1/36, so it follows that the probability of not rolling
a 12 is 35/36, or 97.2%.

 BIG IDEA

We must calculate the probability of a thing happening to correctly calculate the probability of it not
happening.

 TERM TO KNOW

Inverse
The probability of something not happening, or one minus the probability of it happening.

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 45
 2. One Event Happening on Multiple Tries

We can now return to the dilemma we faced earlier, of calculating the probability of winning a game of Three
Card Monte at least once on two tries. In this case, we win if either we pick the Queen the first time or the
second time.

How do we calculate this probability? We can’t multiply probabilities, as we did earlier, because we aren’t
worried about drawing a Queen both times. We also can’t simply add the probabilities together, or we’d get to
100% on three tries, and while we’re likely to win, it’s not 100% certain.

In fact, there is an easy—but not so obvious—solution to this problem. We calculate the probability of not getting
a Queen on either draw, and then subtract that answer from 1. Remember that the probability of all possible
outcomes must sum to 1, so P(Queen) + P(not Queen) = 1.

Again, we’ll assume the dealer is playing a straight game (which is a dangerous assumption in real life!). The
probability of losing the first game is 2/3, or .667. The probability of losing the second game is the same. Using
the formula we learned earlier for two events both occurring, we multiple .667 by .667 and get .445. Remember,
this is the probability of not winning either game, but we can calculate the probability of winning at least once
by subtracting this figure from 1, and get .555. We have a 55.5% chance of winning. We can only arrive at this by
combining two of the formulas we’ve learned, the probability of an event not happening and the probability of
multiple events occurring.

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 46
  WATCH

View a demonstration of this example.

 TRY IT

You can use the same formula for determining the probability of winning at least once on three tries, four
tries, or five. Even a hundred! Use the exponential formula on a calculator to speed this up, with n being the
number of tries. You will notice that no matter how high n is, you will never quite reach 100%.

 FORMULA TO KNOW

3. The Probability of Winning the Lottery

Let’s look at another complex problem: winning the lottery. You know the probability of winning is low, but how
do you calculate the odds?

Let’s say we have a simple lottery where we pick three numbers between 1 and 30. You have to get all three
numbers right to win the jackpot. The balls are drawn from one machine without replacement, meaning that
these are dependent events where no number can be drawn twice.

We can use the formula for multiple events to calculate the probability of getting the five numbers right, keeping
in mind that subsequent draws will change the total distribution of possibilities:

1/30 * 1/29 * 1/28

This comes to 0.00004, or 0.004%. That’s less than one percent of one percent, but if the cost is reasonable
and the reward is large enough, we might go for it.
But wait! We calculated the probability for getting each number right, but forgot a crucial detail. We do not have
to get the numbers in the right order. Say we pick the numbers 7, 11, and 22. Our calculation above shows the
probability of the numbers being drawn in that order. But they could be drawn in the order 11, 22, and 7. Or 22,
7, and 11. We win in any of these combinations. How do we calculate the real probability?

Imagine yourself listening as they draw the numbers. On the first draw, any of the three numbers out of the 30
balls will keep you in the game. On the second draw, there will be two balls out of 29 that will keep you in the
game. (We are only calculating the chances of winning; if the first ball is a miss, you are already out of the
game.) On the last draw, there are 28 balls and only one with your lucky number. So, we would calculate it like
this:

3/30 * 2/29 * 1/28

The chances are now a much more encouraging one in 4,060. But before you run to the corner store to buy a

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 47
 ticket, remember that actual lotteries have five or six draws and between fifty and sixty numbers. You can use
the same formulas to determine the probability of winning, but most calculators aren’t up to the task!

EXAMPLE The multistate Powerball requires the player to pick five numbers between 1 and 69 and one
more number (the eponymous powerball) between 1 and 26, which is independent of the first five. Imagine,
even if you get the first five numbers right (which is about one in 11 million), you still have a 25 in 26 chance
of losing! Fortunately for Powerball players, there is still a prize for matching all but the powerball, but it is
measured in thousands instead of in millions.

4. The Birthday Problem

The birthday problem is a classic problem that measures the probability of two people having the same
birthday. Conceptually, it shows how coincidences can be more common or likely than we first think.

The probability that two people have the same birthday is 1 in 365 (or 4 in 1461, if you consider leap year,
February 29, but we will be ignoring leap year) for .0027, or 0.27%. That is about one fourth of one percent.
Note that the reason this is only 1/365 and not 2/365 is because the 1 here represents a single event of
matching, not the number of birthdays.

Now, what are the chances that in a group of three people, any two will have the same birthday? This might be
perplexing until you remember the trick of calculating the probability of something not happening: calculate the
probability that it happens, then subtract it from 1. We treat each person having a unique birthday (not one they
share with anyone else in the group) as an independent event. So, for three people, we can see the second
person has a 364/365 (99.7%) chance of having a unique birthday, and the third has a 363/365 (99.4%) chance
of having a unique birthday. The chances of both of those occurring is 99.1%. Subtracted from 1, we have a 0.9%
chance of two of those three people having the same birthday.

Obviously, the probability will continue to drop as we add people to the group. The question is, how many
people must be in the party before the probability of two having the same birthday is better than .5? Better than
.75?

 TRY IT

See if you can solve this problem. You can use a calculator, but a spreadsheet is even better.

Check Your Answer +

The probability is slightly higher than 50% when you have 23 people in the group, and better than 75%
when there are 32 people in the group. Is this number smaller than you expected?

The birthday problem is about coincidences. Conceptually, you can use the same reasoning for answering the
question, “What is the chance?” when any coincidence occurs. Perhaps you meet a stranger with the same last
name, or find a book at a used bookstore with the inscription “Property of XYZ,” and you happen to know the
person XYZ. In any of these cases, the probability seems much more remote if you focus on the one outcome.

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 48
 It is unlikely that two people you know will both have the same birthday, but quite likely that any two random
colleagues will have the same birthday. It is unlikely that the person in the next seat on the next flight will have
the same last name as you, but perhaps likely (depending on your name!) that over a lifetime of trips and
encountering strangers, you will meet one or two with your name. The chances of finding a book at Half Price
Books this afternoon that used to belong to Xaviar are indeed long, but if you frequent used bookstores and
have friends who like the same books, the chances of eventually experiencing a coincidence like that are not
so long. The birthday problem (and coincidences in general) compel us to remember the big picture and ask
ourselves, “What is the chance of this never happening?”, to have more context. This doesn’t mean we can’t
enjoy a coincidence and tweet about it!

 SUMMARY

In this lesson, you learned how to calculate more complicated cases involving probability. For example,
the probability of something not happening is easy to calculate by subtracting the probability of it
happening from 1. This is necessary to calculate things like one event happening on multiple tries.
Knowing how to apply multiple rules is necessary for more complex problems like the probability of
winning the lottery, or determining the likelihood of coincidences in the birthday problem.

Source: THIS CONTENT HAS BEEN ADAPTED FROM Introduction to Logic and Critical Thinking.

 TERMS TO KNOW

Inverse
The probability of something not happening, or one minus the probability of it happening.

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 49
 Probability and Statistics
by Sophia

 WHAT'S COVERED

In this lesson, you will learn more about statistical generalizations. Specifically, this lesson will cover:
1. Weighting
2. Reference Class
3. Base Rate

1. Weighting
We can calculate the probability of drawing any card from a deck knowing only the number of total cards and
the distribution of different kinds of cards. For each kind of card, ignoring the suit, there will always be a 4/52
chance, or about a .07 probability, that we will draw that card, and a 1/52 chance, or about .01 probability, if we
take both suit and kind into account. If we are flipping a fair coin (a coin that is balanced on both sides), there is
always a 1/2 chance, or .5 probability, that the coin will land heads, and the exact same probability that it will
land tails. These probabilities are calculated independently of how events have turned out in the past. They
hold purely by virtue of the facts about the cards and facts about the coin.

However, we often say things like:

The Red Sox have a 60% chance of winning this year.

Politician X has an 85% chance of winning the primary election this year.

“Winning” is actually a binary outcome event: you either win or you don’t. In probabilistic terms, these outcomes
both have a .5 probability of success, like flipping a coin. However, then what does it mean to have a 60% or
85% chance of winning? Even though the objective probabilistic generalizations are in conflict with these other
generalizations, the latter are not nonsensical statements.
Recall we said earlier that statistics and probability are closely related. In fact, probability theory underlies
statistics; it’s theoretical math, whereas statistics is applied math. Another way of thinking about it is that
probability is an objective measure for predicting future events based purely on the facts of the game, while
statistics involves tracking event likelihood on the basis of the data (the outcome of past games and real-world
factors that can influence the outcome of a game) we have already observed. Statistics then might be summed
up as the question: From what we have learned about the probability of events that have already happened,
what is the probability of the events to occur again?

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 50
 When we make a statement about a 60% or 85% chance of winning, we don’t base the statement solely on the
(objective) probability of winning and the parameters of baseball (like, how many possible events and outcomes
there are), but rather on the chance of winning, given what we know so far about our team or the candidate’s
performance. We generalize from the data we’ve observed so far.

Weighting is when we assign more or less probability to a particular event, either because the event is more or
less important to the likelihood of the outcome or because we’ve determined that the event is more or less
likely based on the data we’ve observed. We might weigh our team’s objective chances at success because
they’ve recently played several successful games or gained a superstar player.

Why would we use weighting when we calculate the probability of our team’s success? Well, because winning a
sports game is not random in the way that winning at drawing cards or flipping a coin is random. Winning a
sports game is a complex matter depending on several connected causal factors, like the physical and mental
states of the players of both teams on game day, whether there are any injuries, the weather, whether the game
is played at home or away, and potentially a host of other factors. All these auxiliary factors conspire to make
the question of whether the team will win more than a matter of mere 50/50 odds.

Determining whether and how to properly weight your data is a complex question. This is why statisticians often
love sports. It’s often a matter of looking at a lot of data: win records for both teams, the records of each starting
pitcher, who has home field advantage, and other factors. In elections, it involves current polling and historical
trends, such as people rejecting incumbents if the economy is sluggish. Statisticians can differ and debate on
which factors matter the most and how to apply them, but generally do so in good faith.

You can calculate the probability of a pro baseball team winning a game based on records and player stats, but
can you calculate the probability of having a good time at a game? Suppose your happiness is not in any way
dependent on the outcome of the game; it has more to do with the weather, the lines, the behavior of the fans
around you, the quality of the food, etc. This raises a number of questions about which things to measure, how
to measure/quantify them, and how to aggregate them into an “index of enjoyment” that can be calculated, if it
even can be calculated.

Numbers can also be used without the careful statistical analysis, slapped onto a claim to give it an air of
certainty when it is really speculation or wishful thinking. As a critical thinker, you should know how to interpret
and evaluate such claims of probability. Are they based on sound mathematical principles, or is someone just
attaching a number so their claim sounds more precise?

 BIG IDEA

Statistical statements are only as reliable as the data we use to calculate them.
Remember how we evaluate statistical and probabilistic conclusions.

Inductive arguments, by nature, are not claims of certainty, and are thus not weak if their conclusions prove
wrong. However, over time you might give more or less credence to a source based on their record. For
example, if a website predicts the outcome of every baseball game during the season, the teams with a
“60% chance of winning” should win 60% of the time.
We can assess the trustworthiness of the premises in an inductive argument and the reasoning that leads
from the premises to the conclusion. In this case, this means we investigate the data that backs the claims

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 51
 and the methodology used to draw predictions from that data. You can ask if the data is reliable, and if the
methodology is rational. For example, a political prediction based on polls that were conducted online by a
partisan news source should make you skeptical. In that case, the data is suspect. For another example, if a
sports outcome is simply a measure of their winning percentage, you might observe that the formula does
not take their opponent into account, and dismiss it as a weak inductive argument. In that case, the data is
solid (presumably they have counted wins and losses accurately), but the methodology is flawed. If the data
and methodology aren’t easy to find, that is an indication that the claim is inductively weak and perhaps not
made in good faith.

 DID YOU KNOW

Poll analyst and statistician Nate Silver gained notoriety after projecting presidential election results in 2008
and 2012 that were consistent with the outcome. His projections for the election weren’t way off in 2016;
however, the outcome judged to be the more likely (Clinton winning) actually did not occur. As a result, he
was ridiculed by the press and pundits. But really, Silver’s projections followed the same reasoning and by
his own admission, analysis can only be as good as the polls themselves. Further, it’s important to
remember that even if one outcome is more probable than another, given the inherent uncertainty, the
minority outcome can never fully be ruled out as a possibility.

 TRY IT

Find a website that makes a statistical generalization about a sporting event or election. What is the source
of the data, and how easy is it to find an explanation of the data and methodology they use to generate the
generalization?

 TERM TO KNOW

Weighting
When we assign more or less probability to a particular event, either because the event is more or less
important to the likelihood of the outcome or because we’ve determined that the event is more or less
likely based on the data we’ve observed.

2. Reference Class
Let’s look at another way the data someone uses to calculate probabilities can make a difference in how we
interpret the results.

Say Alison is applying to colleges and is considering a school which has a very low 5% acceptance rate. One in
20 students who apply will be accepted, and the other 19 will not. Alison can simply subtract the probability of
getting in (.05) from 1 to see the possibility of not being accepted is 0.95, or 95%.

Of course, this is not simply a random event. Alison’s chances may be much higher or lower depending on her
qualifications. For example, if she is the child or grandchild of a graduate of that particular school, her chances

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 52
 are boosted because many colleges favor “legacy” registrants. In fact, they become an astonishing 33%,
meaning Alison’s chances of getting rejected are a much less daunting 67%.

All probabilities have a reference class, the space of possible events that the outcome of interest is sampled
from. The reference class for drawing a Queen from a deck of cards is, of course, the other cards in the deck.
But which reference class to use often requires careful thinking. In calculating the probability of being struck by
lightning, for example, we should know if the person is a storm chaser, a telephone repairman, or has another
high-risk job, especially if we are giving that person life insurance. In the example above, we saw that Alison’s
reference class as a legacy student makes a big difference in estimating her chances of getting into her reach
school.

For another example, the probability of becoming a billionaire (as of 2021) was calculated to be about one in
600,000, which you might think means that those individuals are incredibly enterprising and clever. But a closer
look into that statistic reveals that well over half of that sample were people born into wealthy families. They
didn’t work hard and go from “rags to riches.” If we use an appropriate reference class to calculate the real
probability of “making it big” in America, the figure changes drastically. In reality, if you come from a working-
class or middle-class family, your chances are closer to 1 in 75 million, whereas if you’re born wealthy, your
chances are 1 in 72,000.

 THINK ABOUT IT

How does factoring in different reference classes in these examples change your perception of getting into
an Ivy League school? Of being a “self-made” billionaire?

 TERM TO KNOW

Reference Class
The field of possibilities in which an event might occur, used to calculate probability.

3. Base Rate
Consider the following scenario. A person goes in for their annual physical exam and tests positive for colon
cancer. It’s a simple screening done on a stool sample and is 80% accurate. That is, in the case of those who
really do have colon cancer, the test will detect the cancer 80% of the time and miss it 20% of the time. The test
will also misdiagnose those who don’t actually have colon cancer 20% of the time. This is approximately correct
for stool screenings.

What is the probability that this person does have colon cancer? The easy guess is 80%, because the test is
right 80% of the time. But this is inaccurate. To understand why, and calculate the right answer, we need to
know the base rate, or prior probability, of getting colon cancer. Intuitively, base rate is the probability of the
target outcome prior to the test. In this case, we can’t say the probability of the person having colon cancer until
we calculate the base rate, or the chances of anyone having colon cancer in the first place. This is one way that
we can weight a probability; by using the prior probability that the event will occur, we can better predict
whether it will occur in the given instance.

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 53
 The base rate in this example is the rate of colon cancer, which is about 4% (it is a bit higher, but we will keep
this simple). To apply the base rate to this calculation, we then have to consider the full range of possibilities
across the population.

We can do this by imagining 100 people getting the test. We know that four of them actually have colon cancer,
and 96 will not. Of those 96, 20%, or about 19, will get a positive test. Of the four who do have colon cancer,
three will get a positive test, and one will get a negative test.

We can construct a table similar to the truth tables we saw earlier to see the full range of possibilities. In this
case, we assign a numeric value to each row, with the number of people in 100 who have that combination of
outcomes. We can thus see the number in the right column as a percentage.

Has Cancer Has a Positive Test Number in 100

True True 3

False True 19

True False 1

False False 77
So, what are the chances you have colon cancer? Of the 22 in 100 who have a positive test, only 3 will be facing
the grim outcome, meaning only 16% of those with a positive test actually have cancer. In other words, even
with a positive test, you can be cautiously optimistic. The probability that it is a false positive is 84%.

The base rate is extremely important to making medical decisions. For another example, let’s look at prostate
cancer. While prostate cancer is a serious risk for people with prostates who are over 55, it is quite rare in those
under fifty, about 0.2% (one fifth of one percent). The general screening for prostate cancer is, like the general
screening for colon cancer, about 80% accurate. It involves a simple blood test. But let’s generate another table
for people under fifty years old who receive a positive result on this blood test. In this case, we will start with
10,000 people, so we can do correct calculations for a low base rate. In this case, only 20 in 10,000 will have
prostate cancer. We can still easily compute the percentage knowing it is the number in the right column
divided by 100.

Has Cancer Has a Positive Test Number in 10,000

True True 16 (0.16%)

False True 1996 (19.96%)

True False 4 (.04%)

False False 7984 (79.84%)

As you can see, even with a positive blood test, there is a less than one percent chance that a person under 50
actually has prostate cancer. This is crucial to know, because the more reliable test for prostate cancer involves
surgery, and would invariably be the next step in diagnosing and treating it. This is one hundred percent
accurate in diagnosing prostate cancer, but the surgery itself can have severe long-term consequences. For
example, about 5% of the people who have the surgery subsequently become incontinent for the rest of their
life.

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 54
 Now let’s generate another table looking at the outcomes for only those with a positive result on the first exam,
presuming all pursue the second exam. The numbers are rounded.

Has Long-Term Effects from

Has Cancer Number in 2012 (from earlier table)
Prostate Surgery

True True 1

False True 100

True False 15

False False 1896

As you can see from the table, a person under fifty going into prostate surgery following a positive blood test is
100 times more likely to suffer from life-altering consequences of the surgery itself than to actually have
prostate cancer. Do you ignore the low risk of actual cancer to prevent the higher risk of long-term
complications from prostate surgery? This is not a purely hypothetical situation, but a genuine dilemma for
doctors and patients.

 TRY IT

Let’s say a hit and run occurs, where a car runs a light, hits another car, then takes off. A single witness says
that the car is a Honda. Now suppose that 8% of cars are Hondas, and that eyewitness accounts describing
the make of car at accidents are accurate 75% of the time.

Should the police be looking for a Honda? You may want to construct a similar table to the ones
+
done above.

If you did construct a table, it would look like this. You can see that the chances of the car being
correctly identified as a Honda are only 6%, and that the chances are much better that it is a different
kind of car and has been misidentified. Therefore, the police should not focus their energy on tracking
down Hondas.

Is a Honda ID’d Witness Number in 100

True True 6

False True 23

True False 2

False False 69

 TERM TO KNOW

Base Rate
The probability of the target outcome prior to the observation of any data.

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 55
  SUMMARY

In this lesson, you learned how calculating real-world probabilities can be complicated. This often
involves weighting, when you assign more or less probability to certain outcomes based on
importance. Other factors that can influence the accuracy and interpretation of a statistical
generalization include considering the reference class used for the generalization, and the base rate of
the outcome, the probability of an outcome prior to the observation of any data. All these factors can
complicate the calculation of a probability, but as critical thinkers, we should be able to interpret and
evaluate statistical generalizations with these factors in mind.

Source: THIS CONTENT HAS BEEN ADAPTED FROM Introduction to Logic and Critical Thinking.

 TERMS TO KNOW

Base Rate
The probability of the target outcome prior to the observation of any data.

Reference Class
The field of possibilities in which an event might occur, used to calculate probability.

Weighting
When we assign more or less probability to a particular event, either because the event is more or less
important to the likelihood of the outcome or because we’ve determined that the event is more or less
likely based on the data we’ve observed.

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 56
 Errors in Reasoning about Probability
by Sophia

 WHAT'S COVERED

In this lesson, you will learn about fallacies and other mistakes people make in probability. Specifically,
this lesson will cover:
1. Small Numbers Fallacy
2. Regression to the Mean Fallacy
3. Gambler’s Fallacy
4. Conjunction Fallacy

1. Small Numbers Fallacy

Suppose a study showed that of the 3,141 counties of the United States, the incidence of kidney cancer was
lowest in counties which are mostly rural, sparsely populated, predominantly Christian, and located in
traditionally Republican states. In fact, this is true. What accounts for this interesting finding? Most people would
be tempted to look for a causal explanation—to look for features of the rural environment that account for the
lower incidence of cancer. However, they would be wrong (in this case) to do so. It is easy to see why once we
consider the counties that have the highest incidence of kidney cancer. These, too, are counties that are mostly
rural, sparsely populated, predominantly Christian, and located in traditionally Republican states!

How is this possible? Crucially, these are not the same counties: there is no county which has both a high rate of
cancer and a low rate of cancer (that would be a contradiction). Rather, if we tracked the rate of cancer in all the
counties in the U.S., we would see that the counties at the top and the bottom of the list have small populations,
and smaller populations are more likely to have extreme results on any statistical measure.

To understand why, let’s consider two friends talking on the phone while driving. They are both using headsets
and being quite safe! Jack is driving on a rural highway and only occasionally sees another car. Jill is driving in
the city and is surrounded by numerous vehicles. Now let’s say they are going to play a game and count how
many of the vehicles they see in the next five minutes are red. After five minutes, they each have a count of the
total vehicles they’ve seen and the number that are red. But Jack has only seen five vehicles, and Jill has seen
500. Now, let’s say the usual rate of red cars on the road is about 10%. Who is more likely to have seen a
number of red cars that is way over 10%?

Keep in mind that we are only interested in the rate of red cars, not the total number of red cars. Think about
Jack. If he saw one red car in the five he counted, that is twice the national average! It is easy to imagine Jack

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 57
 seeing one red car, but for Jill to have seen the same rate of red vehicles, she would have to count 100. That is
much less likely, even with the abundant traffic.

Now, who had a greater chance of seeing a number of red cars much lower than the national average? In this
case, Jack could have easily seen no red cars among the five he counted, which is much lower than the
national average. And again, it is hard to imagine Jill seeing no red cars out of the many she sees. So you can
see how Jack has a better chance of seeing an unusually high percentage of red cars or an unusually low
percentage of red cars, and it might vary even on the same drive, depending on when he starts counting them.

The first lesson here is that smaller samples are more likely to have extremes. In the case of rural counties, the
low numbers lead to outliers on both sides of the same statistical question. This is called the law of small
numbers. It is yet another reason we want our samples to be large and diverse. To make generalizations based
on a small sample in this way is called the small number fallacy.

But there is another lesson here. Consider a statistical generalization like this one:

The incidence of kidney cancer was lowest in counties which are mostly rural, sparsely populated,
predominantly Christian, and located in traditionally Republican states.

Their assumptions probably depend on their demographics. If the person is a rural, Christian Republican, they
might interpret as signs that people like themselves make better choices and live better lifestyles. If the person
is urban, irreligious, and liberal, they may assume that the consequences of crowding and pollution are taking
their toll on people like themselves. To be fair, the question is intentionally presented to lead people to draw
those conclusions. The reality is that counties with small populations are likely to be rural, and rural populations
are likely to be Republican or religious, but none of those details are even relevant to the question. Our brains
are wired to look for causal explanations rather than mathematical explanations, and preferably an explanation
that fits with our ideology. As a critical thinker, it is thus important to ask the right questions. Is the data framed
in the right way? Is the sample large enough? Are the details given due to causation or accidental correlation?

 TERMS TO KNOW

The Law of Small Numbers

The statistical principle that small sample sizes are more likely to lead to extreme outliers on either side
of a statistical question.

Small Numbers Fallacy

The error of looking for causal explanations for extreme outliers that are really simply due to the law of
small numbers.

2. Regression to the Mean Fallacy

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 58
 Suppose that a professional golfer has been on a hot streak, winning tournaments and posting scores that are
several shots under par. Then something happens. The golfer all of a sudden starts playing like average. Not
average like the common duffer on the course, but average for a golf pro. What explains her fall from
greatness? The sports commentators speculate: Could it be that she switched her caddy, or that it is warmer
now than it was when she was on her streak? Or perhaps fame went to her head once she started winning all
those tournaments.

Chances are, none of these are the right explanation because no such explanation is needed. To understand
what is really happening, you must first understand random distribution, which is the way random events occur
over time and the relationship between single events and that distribution.

Let’s say our golfer has an average score of 72. This is “par,” or the number of strokes a first-class golfer scores
on a given course. Obviously, she doesn’t get exactly 72 every round. In fact, she gets an occasional score as
low as 64, and other times goes as high as 80. The law of random distribution would mean that these scores
don’t occur in a predictable way. The low scores and the high scores might “cluster up,” and in this case, the low
scores did cluster up (remember that lower scores are better in golf!).

Unfortunately, the golfer would inevitably regress to the mean. Regression to the mean is a statistical principle
that even if there are a series of outliers (highs and lows) in a variable we’re measuring (as a consequence of
random distribution), over time it will trend toward the statistical average. In sum, the golfer’s hot streak is simply

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 59
 random distribution, and her subsequent return to normal is regression to the mean. But the commentators
want there to be a story, and so will look for causes for any “hot streak” or “slump.”

The regression to the mean fallacy is looking for causal explanations for the change in trends, the ups and
downs that are part of random distributions, or the inevitable regression to the mean. Understanding the laws of
probability helps keep perspective, and in the case of sports, might keep an athlete from asking what she did
wrong, or how to get hot again—and such anxiety is likely to turn into a self-fulfilling prophecy.

 TERMS TO KNOW

Random Distribution
The way that random events occur over time.

Regression to the Mean

A statistical principle that even if there are a series of outliers or unlikely spikes and valleys in
something we’re measuring (as a consequence of random distribution), over time it will trend toward
the statistical average.

Regression to the Mean Fallacy

The error of looking for causal explanations for regression to the mean (or preceding outliers of random
distributions) that are really due to simple statistical rules.

3. Gambler’s Fallacy
The gambler’s fallacy occurs when one thinks that independent, random events can be influenced by past
events. For example, suppose Frank and Erick are playing Rock/Paper/Scissors, the classic hand game where
players have to throw out a random gesture: a clenched fist for rock (which beats scissors), two extended
fingers for scissors (which beats paper), and a flat palm for paper (which beats rock). Presuming neither player
has an unfair advantage, they will each win one third of the time, and one third of the games will be tied (which
happens when both players make the same gesture). Now let’s say Frank wins four games in a row and Erick is
getting frustrated. Another person suggests you make a bet on the next round, and says you can pick the
winner. Do you bet on Frank, who seems to be on a hot streak? Or do you bet on Erick, who is “due” a win?

The answer is that either decision is a gambler’s fallacy, because the past events have no bearing on the next
round. The only safe bet you can make is that over time, each player will win one third of the games.

A misunderstanding of regression to the mean can lead to a kind of gambler’s fallacy. In this case, you might
suppose that Erick will win more of the remaining games, so it all averages out. However, regression to the
mean does not mean that the odds will suddenly tilt in Erick’s favor. It simply means that over time, the win
average will approach the mean. In fact, even with regression to the mean, Frank will likely hold on to his lead
for a while.

 TERM TO KNOW

Gambler’s Fallacy

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 60
 The error of presuming past independent events will affect future independent events.

4. Conjunction Fallacy
Let’s look at one more fallacy that is connected to probabilities and statistical generalizations.

As a college student, Linda was really into the arts. She majored in art history, but also loved to paint and draw,
take photographs, and go to galleries.

Given this information about Linda, which of the following is more probable?

a. Linda is now a CEO.

b. Linda is now a CEO and donates a lot of money to the arts.

If you answered “b,” you are incorrect. This is because of a basic rule of probability, which is that if we have a
conjunct (as we do here), the probability of both conjuncts being true can’t be greater than one conjunct being
true. In this case, the conjuncts are “Linda is now a CEO” and “Linda donates a lot of money to the arts.” The
first statement only requires Linda to be a CEO; the second requires that she be a CEO and that she donate a
lot of money to the arts. Presuming that a conjunction is more likely than one of the conjuncts is called a
conjunction fallacy. Remember, a conjunction is calculated by multiplying the probabilities of the conjuncts,
which gives us smaller probability mass.
Let’s look at an example that is more cut and dried. Which is more likely?

a. I will roll a six on a die.

b. I will roll a six on a die, then roll another six.

It should be fairly simple to see that b is going to be less likely, because the conditions of a have to be met
(rolling a six), then there is a second test (rolling another six). But with human examples, we are more prone to
error. In the example of Linda, we may not pick which one is “more likely,” but which one fits better with the
information we were given and the picture we have of Linda.
All of the example fallacies here are normal, cognitive biases that come from human psychology. As humans,
we look for patterns, for causes, and for explanations. In statistical generalizations, this can lead not just to
drawing conclusions from insufficient data, but drawing further conclusions about incidental or irrelevant details.
In sports and gambling (and other fields), it can mean looking for causes and explanations for hot streaks or
slumps that are really simply due to random distribution and regression to the mean. Gamblers may make
inaccurate predictions based on past events, either presuming that a hot streak will continue or that they are
“due” after a series of setbacks. And in making predictions or judging probabilities, we may jump to conclusions
that fit with our mental pictures and stories. As critical thinkers, it is good to be aware of these fallacies so we
don’t fall prey to them.

 TERM TO KNOW

Conjunction Fallacy

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 61
 The reasoning error where the probability of two events happening is judged to be higher than the
probability of either event happening. This is a fallacy given that the rule for determining the probability
of two events is calculated by multiplying the probability of the two single events together; it yields a
lower probability than either single event. According to probability theory, the conjunction should be
judged to be less likely than either conjunct, but experimental research shows that people don’t behave
that way. This error can result when the additional information makes an outcome fit better with our
preconceived notions.

 SUMMARY

In this lesson, you learned about common misperceptions related to probability and statistical
generalizations. These all involve ignoring or misunderstanding statistical principles. The small
numbers fallacy comes from the law of small numbers, which shows that smaller sample sizes lead to
extreme outliers on either side of any statistical question. Regression to the mean fallacy explains a
random distribution of events and subsequent regression to the mean as being due to other causal
factors. The gambler’s fallacy presumes that past independent events have an effect on future events.
Finally, the conjunction fallacy presumes that a conjunction of two independent events is more likely
than just one of those events. In all of these cases, the psychological impulse to seek out patterns and
explanations leads to errors in judgements about statistical generalizations.

Source: THIS CONTENT HAS BEEN ADAPTED FROM Introduction to Logic and Critical Thinking.

 TERMS TO KNOW

Conjunction Fallacy
The reasoning error where the probability of two events happening is judged to be higher than the
probability of either event happening. This is a fallacy given that the rule for determining the probability of
two events is calculated by multiplying the probability of the two single events together; it yields a lower
probability than either single event. According to probability theory, the conjunction should be judged to
be less likely than either conjunct, but experimental research shows that people don’t behave that way.
This error can result when the additional information makes an outcome fit better with our preconceived
notions.

Gambler’s Fallacy
The error of presuming past independent events will affect future independent events.

Random Distribution
The way that random events occur over time.

Regression to the Mean

A statistical principle that even if there are a series of outliers or unlikely spikes and valleys in something
we’re measuring (as a consequence of random distribution), over time it will trend toward the statistical
average.

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 62
 Regression to the Mean Fallacy
The error of looking for causal explanations for regression to the mean (or preceding outliers of random
distributions) that are really due to simple statistical rules.

Small Numbers Fallacy

The error of looking for causal explanations for extreme outliers that are really simply due to the law of
small numbers.

The Law of Small Numbers

The statistical principle that small sample sizes are more likely to lead to extreme outliers on either side of
a statistical question.

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 63
 Evaluative Language
by Sophia

 WHAT'S COVERED

In this lesson, you will learn about evaluative language, the use of words with connotations to hide
normative premises. Specifically, this lesson will cover:
1. Connotation
2. Rhetoric vs. Reason

1. Connotation
 BEFORE YOU START

Remember that normative arguments are ones that make arguments about what one ought to do, not about
what is observably and factually true. Normative arguments cannot be based entirely on descriptive
statements; there must be at least one normative statement among the premises.
We’ve seen that normative arguments sometimes have hidden premises that assert what we ought to do or
how things ought to be. These hidden premises may or may not be fair game depending on how well they
cohere to our (and others’) moral standards.

For example, say a new homeless shelter is proposed for a neighborhood and becomes the subject of
argument among neighbors. Some might argue that the shelter will lower the quality of life for residents of the
neighborhood. The implicit conclusion is that adding a shelter to the neighborhood is bad for the residents. At
least one of the hidden premises here is that unhoused people degrade the quality of life of others. This is not a
fair assumption. It makes an assumption about the lives of unhoused people and the ways they relate to the
people around them. Suppose at the same time, some people supporting the shelter argue that it is a Christian
organization running the shelter and people’s worries are unfounded, as any organization that is Christian is
morally good. However, this makes a further assumption about Christianity, and it is a highly contested
normative assumption. Further, implicitly, this normative claim assumes that this understanding of Christianity is
a value shared by the community, which is not obvious and quite likely inaccurate.

In the arguments of the supporters, the normative assumptions are encoded in the language itself. Explicit
discussions of Christian organizations being good use evaluative language to influence the audience, to
compel them to accept the argument and hence its conclusion. The language used in describing the
organization as “good” and “Christian” makes explicit the claim that Christian organizations are morally better
than non-Christian organizations.

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 64
 Normative or evaluative language stands in direct contrast to descriptive language. Whereas descriptive
language simply describes a state of affairs, without passing judgment (positive or negative), evaluative
language is used to pass some sort of judgment, positive or negative, on something. However, evaluative
language can also be less transparent than simply naming a moral assumption as a premise.

Contrast the following three statements:

Sam is tall.
Sam is good.
Sam is nosy.

“Tall” is a descriptive term since being tall is, in itself, neither a good nor bad thing. Rather, it is a purely
descriptive term that does not pass any sort of judgment, positive or negative, on the fact that Sam is taller than
average. In contrast, “good” is a purely evaluative term, which means that the only thing the word does is make
an evaluation (in this case, a positive evaluation); it doesn’t carry any descriptive content. “Good,” “bad,” “moral,”
and “immoral” are examples of purely evaluative terms.
Many terms are both descriptive and evaluative; we call these thick terms. For example, Sam is nosy. “Nosy” is
in part a descriptive term, stating that Sam takes an interest in other people’s affairs. But it also has a
connotation, or a feeling that is evoked along with the description. In this case, nosy makes a negative
evaluation of Sam, saying that his interest in others is inappropriate. It implies a hidden premise that there is an
appropriate amount of interest people should have in others, and that Sam has crossed that line.

We could re-describe Sam’s nosiness using purely descriptive language: Sam is very curious about other
people’s affairs. Notice that while the phrase “very curious about other people’s affairs” does capture the
descriptive sense of “nosy,” it doesn’t capture the evaluative sense of nosy, since it doesn’t carry with it the
negative connotation that “nosy” does. If Sam is frequently seen peeking out of his curtains, one might say,
“Sam keeps a close eye on the neighborhood,” which again mixes a description of Sam’s behavior, but now with
evaluative language that presents the behavior in a positive way. It has a hidden premise that neighbors should
look out for each other, or at least that such behavior makes you a good neighbor.

 TRY IT

English has a lot of words which mean the same thing descriptively, but which have different connotations.
They are not truly synonyms because of these nuanced differences. See the examples below.

Sam is unique.
Sam is eccentric.
Sam is odd.

What assumptions can you make about these? What hidden premises are conveyed by the
+
statements?

These all mean that Sam has different interests or behaviors than other people in Sam’s peer group.
The first has a positive connotation, meaning that Sam’s differences make Sam compelling. The second
statement is more or less neutral. The third implies that Sam’s behaviors make Sam an undesirable

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 65
 companion, or at least awkward to be with. The same words might describe the exact same behavior,
depending on the person making the statement.

Sam is confident.
Sam is self-assured.
Sam is arrogant.

What assumptions can you make about these? What hidden premises are conveyed by the
statements? +

Again, the first statement has a positive connotation with the unstated premise that the behaviors of
self-assuredness are desirable. The second statement is neutral. The third has a negative connotation,
with the premise that self-assuredness is not desirable. Notable here is that the words people choose
to describe Sam may depend on characteristics that have nothing to do with Sam’s behavior, like Sam’s
age, race, or gender.

 BRAINSTORM

Can you come up with other lists of three words that describe the same thing but have different meanings
due to connotations? Specifically, ones where there is one with a positive connotation, one with a negative
connotation, and one that is neutral?
Evaluative language is rife in our society, perhaps especially so in political discourse. This isn’t surprising since
by using evaluative language to describe certain persons, actions, or events, we can influence how people
understand and interpret the world. If you can get a person to think of someone or some state of affairs in terms
of a positively or negatively evaluative term, chances are you will be able to influence their evaluation of that
person or state of affairs. That is one of the rhetorical uses of evaluative language.

However, evaluative language can also be slippery. It attempts to sneak premises by us, to influence us without
being transparent. This leads into an important consideration for critical thinking, the use of persuasive
language versus reason.

 TERMS TO KNOW

Evaluative Language
Words and phrases that are normative or make subjective judgments.

Thick Term
A word that has both descriptive and evaluative meaning.

Connotation
An implied evaluative meaning to a word that is also descriptive.

2. Rhetoric vs. Reason

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 66
 Non-rational persuasion is persuasion that appeals to our emotions rather than our reasoning. Such persuasion
is ubiquitous in our current culture. Advertising, political discourse, and even social media posts will often lean
heavily on non-rational persuasion and evaluative language. The field of rhetoric is a discipline that explores
the full range of persuasive discourse.

Both persuasion and logic are largely about arguments, but there is a key difference between trying to convince
somebody based on facts and evidence and trying to convince them by emotional ploys. This led to historic
conflict between logicians, who wanted arguments to be entirely based on reason, and rhetoricians, who would
use other appeals such as appeals to emotion or pride. Aristotle played a crucial role in trying to bridge this gap
between rhetoric and logic. He wrote several important texts on the role of logic in strong persuasive
arguments.

Rhetoric can and often does include logic; it is one of the three key elements of persuasion, along with
emotions and personal standing. But rhetoric can largely rest on emotional ploys that are not based on reason.
If a magazine ad for insurance repeatedly uses the word “trust” (especially combined with heart-tugging
imagery), that association might sink in, even subconsciously. Beer commercials use a more visual language,
showing people having fun without telling you anything about the beer itself (other than it “tastes great,” or
some other entirely subjective comment). The use of emotionally loaded terms and images are hidden
persuaders, as journalist and social critic Vance Packard described them—the subtle ways advertisers (and
other industries) use non-rational elements to influence our behavior. These are extremely effective and often
more successful than facts.

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 67
 Of course, rhetoric can be used for better ends than selling beer. Throughout history you can find powerful and
persuasive rhetoric that is based more on feeling than reason and is used to impel people to do the right thing:
Mahatma Gandhi and Martin Luther King come to mind. Some of this is because there are limitations to reason
as a tool. You can convince a person something is true, but you may not be able to convince them to care that it
is true and to take action without using other rhetorical strategies. In fact, appealing to emotions is crucial to
even getting people to listen to you in the first place. In Aristotle’s time, there was debate over whether
persuasion should be used at all, or if arguments should appeal to pure reason.

 THINK ABOUT IT

If you can convince somebody to do the right thing, does it matter if you do it in deceptive or manipulative
ways?
You may fall on the side of the rhetoricians, as most people do. But as critical thinkers, we need to be analytical
about the arguments we are presented with, identifying the hidden persuaders and evaluative language that is
attempting to guide us toward a conclusion. We identify the hidden premises and ask whether or not we accept
them. If engaging in persuasion, we have to decide where we stand ideologically on the use of persuasion or
use of pure reason to achieve our goals. Either way, we should make our arguments in good faith.

 TERMS TO KNOW

Non-Rational
Words or actions based on emotions instead of reason.

Rhetoric
The art of persuasion.

Hidden Persuaders
The use of images, evaluative language, and other techniques to influence people without their direct
awareness.

 SUMMARY

In this lesson, you learned about evaluative language. Evaluative language uses words with good or
bad connotations to conceal hidden premises and assumptions; these can be quite effective in making
an argument but are not based on reason. This raises the question of rhetoric vs. reason, and whether
we should use only reason to convince people of our position, or if we should use any persuasive
techniques we can. In truth, emotional ploys can be used for trivial things like selling beer but have also
been employed by major historical figures to foment positive change. Ultimately, as critical thinkers, the
important thing is to analyze arguments for hidden persuaders and to make our own arguments in
good faith.

Source: SOURCE: THIS CONTENT HAS BEEN ADAPTED FROM Ethics for A-Level and Introduction to
Philosophy: Ethics.

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 68
 REFERENCES

Packard, V. (1957). The Hidden Persuaders. New York: The David McKay Company.

 TERMS TO KNOW

Connotation
An implied evaluative meaning to a word that is also descriptive.

Evaluative Language
Words and phrases that are normative or make subjective judgments.

Hidden Persuaders
The use of images, evaluative language, and other techniques to influence people without their direct
awareness.

Non-Rational
Words or actions based on emotions instead of reason.

Rhetoric
The art of persuasion.

Thick Term
A word that has both descriptive and evaluative meaning.

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 69
 Moral Truth (Cognitivism and Noncognitivism)
by Sophia

 WHAT'S COVERED

In this lesson, you will learn about moral truth and the complications of making moral claims.
Specifically, this lesson will cover:
1. Moral Truth
2. Cognitivism and Non-cognitivism
2a. Moral Cognitivism
2b. Non-Cognitivism
2c. Prescriptivism
3. Fallacious Justifications for Moral Beliefs
3a. Legality
3b. Prudence
3c. Science
3d. Moral Diversity
4. Moral Truth and Critical Thinking

1. Moral Truth
In the last two units, we’ve looked extensively at evaluating arguments or positions for validity and strength,
soundness, and cogency. We can check the argument’s internal structure to assure that the premises support
the conclusion. If the argument is fact-based, we can evaluate the premises by comparing them to the world
around us. That is, we can simply fact-check.

But what about the normative arguments, those arguments based on values (often implicit but not stated
outright). These introduce a number of difficult questions. Whether those statements are true or not depends on
our value system: how we interpret “good” and “bad,” “right” and “wrong,” and other value-laden terms. These
lead us down a philosophical rabbit hole.

When we say something is “morally good,” what do we mean?

If the claim that “[activity] is morally wrong” is true, what makes it true?
How do we fact-check moral claims?
Can “ought” statements be true or false?

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 70
 Think of a time when you disagreed with someone about the right thing to do. Maybe it was a friend, family
member, celebrity, author, or political figure. You may have felt very strongly that X is obviously the right thing to
do, the better course of action, or merely the lesser of two evils. The person you were disagreeing with might
have felt similarly, and perhaps provided reasons for their position as well. Both of you made claims about
morality. You each believed that your own position was correct or true. But are these claims about morality true
or false in the same way that historical and mathematical facts are true or false? “George Washington was the
thirteenth president of the United States of America” is a false historical claim because George Washington was
not the thirteenth president of the United States of America. Why is this historical claim false? Because it goes
against reality.
The question is whether there is such a reality for morals. Are there moral facts that hold true regardless of what
we think about them? Are there moral facts that are true by virtue of some mind-independent moral reality? A
mind-independent fact is true even if nobody is aware of it. A concept is mind-independent if and only if its truth
or falsity depends only on facts about the world

EXAMPLE The presence of atoms was true for millennia before humans knew they existed; the
existence of atoms is a mind-independent fact. The rate of acceleration caused by gravity is mind-
independent—it is dependent only on facts about the world (i.e., the mass of the nearest celestial body). On
the other hand, whether or not food is “tasty” is mind-dependent. The tastiness of a particular food, say
curry, will be dependent on the mind and the tastebuds eating the food.

 THINK ABOUT IT

Is there a moral truth you believe in that others do not? What evidence or argument can you make that
supports it?

 TERM TO KNOW

Mind-Independent
Objective reality that does not require human thought to be true.

2. Cognitivism and Non-cognitivism

For the purposes of this class, we cannot consider the full range of ethical theories or lines of questioning for
discussing morality. However, we must learn some basic terminology to understand why moral reasoning is
different from other forms of reasoning. In fact, not all people agree that “moral reasoning” is even possible.
Some would say that “moral reasoning” is a contradiction in terms!

Let’s consider three basic ways of thinking about moral statements.

2a. Moral Cognitivism

Moral cognitivism is a philosophical approach that suggests that when we make moral claims of the form
“murder is wrong” or “helping others is right,” our claims can be true or false; they have truth value.

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 71
 Some varieties of moral cognitivism hold that moral truths are mind-independent and others do not. Here’s an
example of a kind of moral cognitivism that is mind-independent. Suppose that statements like “murder is
wrong” are true or false depending on facts about the world. It doesn't matter whether there are people in the
world (murder is wrong on Mars too), or how those people perceive murder. The morality of murder is
determined and immutable just like other facts about the world, like gravity.

On the other hand, if moral cognitivism is mind-dependent, then in a world without humans, “murder is wrong”
is a meaningless statement. A person can only say that “murder is wrong” is a moral truth insofar as it is
supported by what the utterer believes. If moral statements are mind-dependent, then cultures and people’s
beliefs are instrumental in determining the truth value of the statement.

Stating that a moral claim like "murder is wrong" is true may seem obvious, but many philosophers consider this
stance to be incorrect. We will explore alternative theories below in this section.

According to the cognitivist, what makes our moral statements true or false is whether or not they accurately
describe moral aspects of the world. Thus, the cognitivist views our moral language as essentially descriptive in
nature; we try to describe genuinely moral features of the world, and our moral claims are true when our
descriptions are accurate and false when they are inaccurate. Cognitivism may also be expressed as a personal
belief, stating something like, “I believe murder is wrong,” or “Everybody deserves a second chance.” However,
there is no easy distinction between a personal claim and a more descriptive claim; the implication of a personal
belief is that it accurately describes the world. It doesn’t make much sense to say, “I believe child abuse is
wrong,” and not mean it as a universal standard, that anybody abusing any child is wrong. In a philosophy class,
you may pursue different threads of moral cognitivism, but for the purposes of this class, we will treat it as a
single idea that moral statements can be true because (a) the person making the claim believes it, and (b) the
person making the claim feels it describes the best moral truth for the world.

Cognitivism is the commonsense position when it comes to moral language. In constructing an argument, you
would make normative statements that are presented as truths, and which presumably have little disagreement.

 TERM TO KNOW

Moral Cognitivism
The belief that moral statements have truth value.

2b. Non-Cognitivism
As you might guess, non-cognitivism is the opposite of cognitivism. According to the non-cognitivist,
statements such as “murder is wrong” do not describe any actual features of the world but are simply an
attitude or feeling—in this case, disgust and anger. They might be likened to a cry of despair. “Aaaah!” Feelings
can’t be true or false. They are not falsifiable or defeasible. Responding to a cry of despair by saying “true” or
“false” would not make sense, and so (says the non-cognitivist) it makes no sense to respond to moral
statements as true or false.

The non-cognitivist thus suggests a fairly radical understanding of our common views regarding what moral
utterances mean and how moral discourse works. As you might guess from the names “cognitivism” and “non-
cognitivism,” these positions are essentially about whether or not moral statements are based on reason or

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 72
 pure emotion. If they are based on pure emotion, they cannot be true or false, any more than a laugh or a cry
can be true or false.

 TERM TO KNOW

Moral Non-Cognitivism
The belief that moral statements do not have truth value, but are just feelings.

2c. Prescriptivism
A compromise of the cognitivist and non-cognitivist positions on moral statements is prescriptivism. A
prescriptivist describes moral statements as feelings, like the non-cognitivists, but also presumes that a moral
statement is also a declaration of how others should behave. To a prescriptivist, “murder is wrong” is not just an
expression of disgust, but it indicates a normative belief. It thus finds a middle ground where a moral statement
can be true, to the extent that it implies a sentiment like, “You should not commit murder.” However, it does not
imply objective moral truths.

Crucially, prescriptivism gives us a way to introduce normative statements to arguments, as well as a way of
analyzing moral statements. We will pursue this more in the next tutorial.

 THINK ABOUT IT

Where do you fall in these three categories? Do you believe that moral statements are declarative or simply
feelings?

 TERM TO KNOW

Moral Prescriptivism
The compromise between cognitivism and non-cognitivism that accepts moral statements as essentially
feelings, but we can extrapolate normative statements that do have truth value.

3. Fallacious Justifications for Moral Beliefs

As we attempt to “prove” that our moral statements are true or false, it is tempting to use empirical evidence
that is objectively measurable or observable. However, empirical evidence does not prove a moral statement is
true or false.

3a. Legality
Whether or not something is against the law is not evidence it is moral or immoral. For example, many things
are immoral but legal—like cheating on your spouse, or lying (in most cases, unless it is outright fraud). Other
activities may be moral but illegal, like helping enslaved people escape was in some states before 1863.
Therefore, in discussions about moral duty, be wary of talking about a thing being legal or illegal as proof of
moral truth or falsehood.

 THINK ABOUT IT

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 73
 Can you think of any examples now of things that are moral duty but illegal?

3b. Prudence
Whether or not something is prudent is also not evidence a thing is moral or immoral. Prudence relates to our
practical or pragmatic reasons for doing things. For example, imagine a town where one big factory is the
primary employer. They are discovered to be dumping pollution into the river. People might argue that to shut
down the factory would be devastating economically to the town. This might be true but does not mean the
factory’s actions are moral or immoral. There are frequent appeals to practicality on moral questions that do not
show proof either way for a moral claim.

3c. Science
Whether or not there is a scientific basis for a particular behavior does not mean it is moral or even morally
neutral. For example, supporters of meat-eating often point to our incisor teeth, or to evolutionary evidence that
meat-eating was crucial to the development of human brains. This shows that it is natural for us to eat meat, and
may even describe our desire to eat meat, but does not tell us how we morally ought to behave. Indeed, moral
reasoning is largely about curbing our instinctive behaviors.

3d. Moral Diversity

Arguing that people disagree about a moral issue is not an argument against there being any real answer to the
question of whether there is a moral truth. Imagine that a group of friends are arguing about which country has
won the most Olympic gold medals. Max says China, Alastair says the U.S., and Dinh says the UK. There is
disagreement, but does this mean that there is not an answer to the question of “which country has won the
most Olympic gold medals?” No! We cannot move from the fact that people disagree to the conclusion that
there is no answer.

Now consider a parallel argument that we hear far too often. Imagine that you and your friends are discussing
whether eating meat is morally acceptable. Some say yes, the others say no. Each of you cite how different
cultures have different views on eating meat. Does this fact—that there is disagreement—mean that there is no
answer to the question of whether eating meat is morally acceptable? Again, the answer is no. That answer did
not follow in the Olympic case, and it does not follow in the moral one either. So, just because different people
have different moral views, this does not show, by itself, that there is no moral truth and no answer to the
question.

4. Moral Truth and Critical Thinking

When you evaluate what is right and wrong, it can be tempting and comforting to spend time simply defending
your initial views; few people would come to a debate about vegetarianism, or abortion, without some pre-
existing belief. If you are open-minded in your ethical approach, then you need not reject everything you
currently believe, but you should see these beliefs as starting points, or base camps, from which your inquiry
commences. For example, why do you think that eating animals is OK, or that abortion is wrong? If you think
that giving to charity is good, what does “good” mean?

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 74
 To be a good critical thinker requires intellectual respect for other beliefs. If you think that a particular position is
obviously false, perhaps take this reaction as a red flag, as it may suggest that you have missed some important
step of an argument—ask yourself why someone, presumably just as intellectually proficient as yourself, might
have once accepted that position. If you are thinking well as an ethicist, then you are likely to have good
reasons for your views, and be prepared to rethink those views where you cannot find such good reasons. In
virtue of this, you are providing justification for the beliefs you have.

To be a good critical thinker, whatever beliefs you have, you should ask why you hold those beliefs. For
example, imagine the reason that you believe it is OK to eat meat is that it tastes good. This is not a particularly
good reason. Presumably your pet cat, or your neighbor, or your dead aunt would also “taste good,” but in
these cases, the “taste good” justification seems totally inadequate!

The point is that there are good and bad reasons for your beliefs. You need to be authentic with yourself and to
ask what you think, using this as a guide to critically analyze the ideas learned and lead yourself to your own
justifiable conclusion.

 SUMMARY

In this lesson, you learned about the complexity of interpreting moral claims as having truth value—that
is, whether or not there is even such a thing as moral truth. The topic is divided between cognitivism
and non-cognitivism. Moral cognitivism is seeing moral statements as based on reason and having
truth value. Non-cognitivism sees moral statements as feelings and not having truth value.
Prescriptivists see moral statements as feelings, but since a moral feeling implies a normative
statement about how people ought to behave, they ultimately have truth value. In considering moral
truths, it is important to avoid fallacious justifications such as citing legality, prudence, science, or
moral diversity. Knowing how to maintain our own beliefs while respecting others that differ is at the
heart of moral truth and critical thinking.

Source: SOURCE: THIS CONTENT HAS BEEN ADAPTED FROM Ethics for A-Level and Introduction to
Philosophy: Ethics.

 TERMS TO KNOW

Mind-Independent
Objective reality that does not require human thought to be true.

Moral Cognitivism
The belief that moral statements have truth value.

Moral Non-Cognitivism
The belief that moral statements do not have truth value, but are just feelings.

Moral Prescriptivism

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 75
 The compromise between cognitivism and non-cognitivism that accepts moral statements as essentially
feelings, but we can extrapolate normative statements that do have truth value.

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 76
 Moral Frameworks
by Sophia

 WHAT'S COVERED

In this lesson, you will learn about frameworks for moral reasoning. Specifically, this lesson will cover:
1. Utilitarianism
1a. Utilitarianism: An Introduction
1b. The Foundations of Bentham’s Utilitarianism
1c. The Structure of Bentham’s Utilitarianism
1d. Mill’s Qualitative Utilitarianism
2. Deontology
3. Virtue Ethics
3a. The Function Argument
3b. Aristotelian Goodness
3c. Developing the Virtues

1. Utilitarianism
In the last tutorial, we saw the difficulty in asserting moral truths, which in turn support most normative
arguments. If we assert that a policy is bad because it impoverishes people, it rests on a moral claim that we
should not make the lives of other people harder, even if we otherwise pursue selfish goals. We saw that some
believe in moral truth, in the sense that a moral statement can have truth value, and that others reject moral
truths, considering them more in the territory of feelings than statements with truth value. Finally, we saw that
there is a kind of compromise in understanding a moral statement to be both a feeling and a claim about how
others should behave. Any of these rely on assumptions about what is “good” to begin with. Let’s explore the
different ways that philosophers have gone about finding what “good” requires.

1a. Utilitarianism: An Introduction

Some things appear to be straightforwardly good for people. Winning the lottery, marrying your true love, or
securing a desired set of qualifications all seem to be examples of events that improve a person’s life. As a
normative ethical theory, utilitarianism suggests that we can decide what is morally right or morally wrong by
weighing which of our future possible actions promotes such goodness in our lives and the lives of people
more generally.

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 77
  TERM TO KNOW

Utilitarianism
The belief that we can decide what is morally right or morally wrong by weighing which of our future
possible actions promotes such goodness in our lives and the lives of people more generally.

1b. The Foundations of Bentham’s Utilitarianism

Jeremy Bentham (1748–1832) was the first of the “classical utilitarians.” Bentham developed his moral theory of
utilitarianism on the foundation of hedonistic thinking. Hedonism is a theory of well-being which argues that
what defines a successful life is directly correlated with the amount of pleasure in that life; no other factors are
relevant. For Bentham, the only thing that determines the value of a life, or indeed the value of an event or
action, is the amount of pleasure contained in that life, or the amount of pleasure produced as a result of that
event or action—this is what makes Bentham fundamentally hedonist.

When first understanding utilitarianism, it is also crucial to understand what is meant by the term “utility.”
Bentham defined it as “[…] that property in any object, whereby it tends to produce benefit, advantage,
pleasure, good, or happiness […] or […] to prevent the happening of mischief, pain, evil, or unhappiness.” Utility
is thus promoted when pleasure is promoted and when unhappiness is avoided. Bentham’s commitment to
hedonism means for him that goodness is just an increase in pleasure, and evil or unhappiness is just an
increase in pain or decrease in pleasure. With this understanding of utility in mind, Bentham commits himself to
the principle of utility.

In effect, this principle simply says that promoting utility, defined in terms of pleasure, is to be approved of and
reducing utility is to be disapproved of. Another way to rephrase this principle of utility is as a requirement to
promote the greatest pleasure for the greatest number of people, in order to act morally.

 TERM TO KNOW

Hedonism
A theory of well-being which argues that what defines a successful life is directly correlated with the
amount of pleasure in that life.

1c. The Structure of Bentham’s Utilitarianism

In addition to being hedonistic, Bentham’s utilitarianism is also:

1. Consequentialist/Teleological
2. Relativist
3. Maximizing
4. Impartial

Bentham’s utilitarianism is consequentialist because the moral value of an action or event is determined entirely
by the consequences of that event. The theory is also described as teleological for the same reason, based on
the Greek word telos that means “end” or “purpose.”
In addition, Bentham’s utilitarianism is relativistic rather than absolutist. Absolutist moral views hold that certain
actions will always be morally wrong irrespective of context or consequences. For example, many campaigning

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 78
 groups suggest that torture is always morally unacceptable. For absolutists, then, the act of torture is absolutely
wrong in all cases and situations. However, a relativist like Bentham cannot hold this type of view because
sometimes the pain involved in torture may lead to the promotion of greater pleasure (or less intense pain)
overall, such as in the case where information gained through torture is used to stop a terrorist atrocity. On this
basis, the Benthamite utilitarian must believe that whether a certain action is right or wrong is always relative to
the situation in which the action takes place.

Bentham’s utilitarianism is maximizing because it does not merely require that pleasure is promoted, but that
the greatest pleasure for the greatest number is secured. Thus, for example, if you gain some pleasure from
spending money on a new book, but that money could have produced more pleasure had it been donated to a
local charity for the homeless, then buying a new book would be morally wrong even though it led to some
pleasure because it did not maximize the total amount of pleasure that was possible in that circumstance.

Finally, Bentham’s utilitarianism is also impartial in the sense that what matters is simply securing the maximum
amount of pleasure for the maximum number of people; the theory does not give special preference regarding
which people are supposed to share in that total pleasure. In the total calculation of pleasure, we are all equal
regardless of our status, behavior or any other social factor.

 TERMS TO KNOW

Consequentialism
The belief that the moral value of an action or event is determined entirely by the consequences of that
event.

Teleology
The belief that moral actions are determined by circumstances and outcomes.

Absolutism
The belief that certain actions will always be morally wrong irrespective of context or consequences.

Relativism
The belief that actions should be morally evaluated in the context and with regard to consequences.

1d. Mill’s Qualitative Utilitarianism

In attempting to refine and improve Bentham’s utilitarianism, John Stuart Mill’s (1806-1873) most substantial
thought was to move away from Bentham’s idea that all that mattered was the quantity of total pleasure.
Instead, Mill thought that the quality of pleasure was also crucial to deciding what is moral.

Bentham’s utilitarianism is quantitative in the sense that all Bentham focuses on is the maximization of
hedonically calculated quantities of total pleasure. All that matters for Bentham is producing pleasure, and the
way this is achieved is unimportant. If playing Fortnite affords you more pleasure than reading Shakespeare,
then Bentham would view your life as going better if you play Fortnite. However, Mill introduces a quality
criterion for pleasure. Mill says that:

It is better to be a human being dissatisfied than a pig satisfied; better to be Socrates dissatisfied than a
fool satisfied. And if the fool, or the pig, is of a different opinion, it is only because they only know their own

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 79
 side of the question.

According to Mill, higher pleasures are worth more than lower pleasures. In Mill's view, higher pleasures are
those pleasures of the intellect brought about via activities like poetry, reading or attending the theater. Lower
pleasures are animalistic and base—pleasures associated with drinking beer, having sex, or lazing on a sun-
lounger. What we should seek to maximize are the higher quality pleasures, even if the total pleasure turns out
to be quantitatively lower as a result.

 THINK ABOUT IT

Do you agree that some pleasures are "higher" or "lower" than others? What is an activity that is a high or
low pleasure for you? Would another person categorize it the same way for themselves?

2. Deontology
Another way of approaching the definition of “good” is to think about one’s moral duties. Moral duties refer to
the idea that one has a set of responsibilities which will govern behavior and decisions.

Specifically, this brings us the idea of a categorical imperative, which has its origins in the writings of Immanuel
Kant (1724-1804). An imperative is an order: clean your room, don’t talk to strangers, etc. The word “categorical”
here means unambiguous and direct—a categorical imperative is a clear rule that applies without ambiguities,
with straightforward directives about what action is good and what is bad.

Some imperatives are situational. For example, “clean your room” only makes sense if the room isn’t clean, if
you care if the room is clean, etc. But a categorical imperative would always be true. “Don’t hurt children” is true
in any situation, so it is categorical, not situational.

The meaning of Kant’s moral categorical imperative is that all moral decisions and actions we make can only be
carried out if they can be taken as universal laws. It is like the golden rule, “Do unto others as you would have
them do unto you,” but taken further, “Do as you would have all people do all the time.”

Importantly, intentions and consequences are not part of the categorical imperative; only the action itself should
be judged. For example, take lying. Most people lie at different times for different reasons—it may be self-
serving, or simply to spare someone’s feelings. But by the categorical imperative, we would have to consider
whether we want "you should lie" to be a universal law, not whether a specific case of lying is justifiable based
on good intentions or desirable consequences. Kant believed that humans have good will, a basic human
condition of wanting to be good that supersedes (usually) other desires, and that this good will would guide
them to seeing the morality of every action irrespective of how they might feel and irrespective of any
consequences. Moreover, this good will would transcend cultural boundaries. Lying, for example, would be
wrong because people would not want "you should lie" to be a universal law.

IN CONTEXT

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 80
 Crucial to Kant’s theory is that people do have an impulse to do good will. But how can he (and we) be
confident that the good will even exists? Consider Mahatma Gandhi’s (1869–1948) nonviolent protest
for Indian independence. He stood peacefully whilst the British police beat him. Here is a case where
there must have been an overwhelming desire to fight back. But he did not. In this type of action, Kant
would claim that we “see” the good will—as he says—“shining like a jewel.” Seeing such resilience in
the face of such awful violence, we are humbled and can recognize good will.

Kant’s theory is deontological, which judges actions based on essential moral rules. This is distinct from
consequentialism (like utilitarianism), which judges actions based on intentions and outcomes. However, Kant
doesn’t assert those moral rules, but gives us a way to use reason to attain them. While there is far more to
Kant’s philosophy than we can cover in this tutorial, through his arguments we have a way to assert a kind of
supreme moral truth, which he calls moral duty. Moral duty is the essence of living by the categorical
imperative. That is, we have decided to do the right thing, which we determine by considering what the world
would be like if everyone did as we did. We do not let any other consequences, material or emotional, affect
that decision.

 HINT

A deontologist is not interested in consequences, but whether the rules we use to make decisions can be
universalized. An action is the right one according to a deontologist if it is what we would recommend all
people do all the time.

 THINK ABOUT IT

Suppose you find a wallet in the street, and it is loaded with cash. It also has an identification card, and
belongs to a celebrity you admire. There are now reasons pulling you in different directions—the material
gain of keeping the cash, the emotional fulfillment of meeting someone you admire when you return the
wallet (maybe even taking a few selfies for social media), etc. To live by the categorical imperative, we
would consider neither of these, only the rule we are setting up. Should people live by the “finders keepers”
rule, or go through efforts to return lost things? Consider the consequences both ways. What would the
categorical imperative have us do?
Kant insists that we act from duty, not in accordance with duty—that is, we should not only do the right thing but
do it for the right reasons. In the case where we find a celebrity’s wallet, we might do the right thing (return the
wallet) to get an emotional reward (the chance to meet someone we admire). If we do it more to meet the
celebrity and just happen to also be doing the right thing, it would be in accordance with moral duty but not
acting in moral duty. Obviously, we would not be living by the categorical imperative if we only return a lost item
when there’s something in it for us!

 TERMS TO KNOW

Categorical Imperative
The idea that our moral actions should be carried out as if they were universal laws.

Good Will

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 81
 In Kantian philosophy, the idea that humans have an instinct toward goodness that can supersede all
other desires.

Deontology
The belief that moral actions are set by universal laws.

Moral Duty
Making decisions based on our moral obligations or responsibilities.

3. Virtue Ethics
The consequentialist and deontologist have different methods for making moral decisions. Virtue ethicists
approach the problem differently, not considering individual decisions or actions but how to become good
people.

3a. The Function Argument

Aristotle (384–322 BC) was a scholar in disciplines such as ethics, metaphysics, biology, and botany, amongst
others. It is fitting, therefore, that his moral philosophy is based around assessing the broad characters of
human beings rather than assessing singular acts in isolation. Indeed, this is what separates Aristotelian virtue
ethics from both utilitarianism and Kantian ethics. Aristotle was a teleologist, a term related to, but not to be
confused with the word “teleology” as it is used in utilitarianism. In utilitarianism and other consequentialist
beliefs, teleological means that the consequences can justify an action. Aristotle’s teleology is more than
justifying a single decision.

This view is essentially that in achieving its function, goal, or end, an object achieves its own good. Every object
has this type of a true function, and so every object has a way of achieving goodness. The telos of a chair, for
example, may be to provide a seat, and a chair is a good chair when it supports the curvature of the human
bottom without collapsing under the strain. Equally, says Aristotle, what makes good sculptors, artists, and
flutists are the successful and appropriate performances of their functions as sculptors, artists, and flutists.

This teleological-based (function and purpose) worldview is the necessary backdrop to understanding
Aristotle’s ethical reasoning. For, just as a chair has a true function or end, so Aristotle believes human beings
have a telos. Aristotle identifies what the good for a human being is by virtue of working out what the function
of a human being is, as per his function argument.

 KEY CONCEPT

Function Argument
1. All objects have a telos.
2. An object is good when it properly secures its telos.

Given the above, hopefully these steps of the argument are clear so far. At this point, Aristotle directs his
thinking towards human beings specifically.
1. The telos of a human being is to reason.

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 82
 2. The good for a human being is, therefore, acting in accordance with reason.

In working out our true function, Aristotle looks to that feature that separates man from other living animals.
According to Aristotle, what separates mankind from the rest of the world is our ability not only to reason
but to act on reason. We achieve goodness when we act in accordance with this true function or telos.

 TERM TO KNOW

Teleology (in Virtue Ethics)

The view that all things (including people) have a purpose and achieve goodness by fulfilling that
purpose. Aristotle’s view was that the purpose of people is to live in accordance with virtue and reason.

3b. Aristotelian Goodness

On the basis of the previous argument, the good life for a human being is achieved when we act in accordance
with our telos. However, rather than leaving the concept of goodness as general and abstract, we can say more
specifically what the good for a human involves. Aristotle uses the Greek term eudaimonia to capture the state
that we experience if we fully achieve a good life. According to Aristotle, eudaimonia is the state that all humans
should aim for, as it is the aim and end of human existence. To reach this state, we must ourselves act in
accordance with reason. Properly understanding what Aristotle means by eudaimonia is crucial to
understanding his virtue ethical moral position.

 HINT

An adherent to virtue ethics believes right actions come from cultivating good and virtuous people. So, a
virtue ethicist will recommend the action that exhibits the virtues rather than vices
Eudaimonia has been variously translated and no perfect translation has yet been identified. While all
translations have their own issues, eudaimonia understood as flourishing is perhaps the most helpful translation
and improves upon a simple translation of happiness. The following example may make this clearer.

IN CONTEXT

Naomi is an extremely talented pianist. Some days, she plays music that simply makes her happy,
perhaps the tune from the Australian television soap opera Neighbours or a rendition of “Twinkle,
Twinkle Little Star.” On other days, she plays complex music such as the supremely difficult Chopin-
Godowsky Études. These performances may also make Naomi happy, but she seems to be flourishing
as a pianist only with the latter performances rather than the former. If we use the language of
function, both performances make Naomi happy, but she fulfills her function as a pianist (and is a good
pianist) only when she flourishes with the works of greater complexity.

Flourishing in life may make us happy, but happiness itself is not necessarily well-aligned with acting in
accordance with our telos. Eudaimonia is secured not as the result of exercising our physical or
animalistic qualities, but as the result of the exercise of our distinctly human rational and cognitive
aspects.

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 83
  TERM TO KNOW

Eudaimonia
The state of achieving a good life in accordance with our purpose; flourishing.

3c. Developing the Virtues

Will Durant (1885–1981) sums up the Aristotelian view by saying that “…we are what we repeatedly do.
Excellence, then, is not an act but a habit.” The quote is often misattributed to Aristotle himself.

It is fairly obvious that we cannot become excellent at something overnight. Making progress in any endeavor is
always a journey that requires both effort and practice over time. Aristotle holds that the same is true for human
beings attempting to develop their virtuous character traits in an attempt to live the good life. You may feel
yourself coming to an Aristotelian virtue ethical view after reading this tutorial and therefore be moved to
become wittier, more courageous, and more generous, but you cannot simply acquire these traits by decision;
rather, you must live these traits in order to develop them.

Cultivating a virtuous character is something that happens by practice. We might know that a brick must go into
a particular place, but we are good builders only when we know how to place that brick properly. Ethical
characters are developed by practical learning and habitual action and not merely by intellectual teaching.

In the end, the virtuous individual will become comfortable in responding to feelings/situations virtuously just as
the good builder becomes comfortable responding to the sight of various tools and a set of plans. A skilled
builder will not need abstract reflection when it comes to knowing how to build a wall properly, nor will a skilled
cyclist need abstract reflection on how to balance his speed correctly as he goes around a corner. Analogously,
a person skilled in the virtues will not need abstract reflection when faced with a situation in which friendliness
and generosity are possibilities; they will simply know on a more intuitive level how to act.

 SUMMARY

In this lesson, you learned about the concept of moral duty, or how we weigh moral decisions.
Utilitarianism is the notion that we make decisions based on their outcomes, and strive to maximize
pleasure for ourselves and others (thus, reducing harm). By contrast, deontology is more absolute,
believing that moral decisions are based on guiding principles regardless of consequences. In virtue
ethics, these decisions are seen as our very purpose in life, to live virtuous lives and achieve
eudaimonia.

Source: SOURCE: THIS CONTENT HAS BEEN ADAPTED FROM Ethics for A-Level.

REFERENCES

Arendt, H. (1994). Eichmann in Jerusalem : a Report on the Banality of Evil. New York: Penguin Books.

Bentham, J. (1830) The Rationale of Reward, p. 206 books.google.co.uk/books?id=6igN9srLgg8C

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 84
 Durant, W. (1926) The Story of Philosophy, p. 87
www.google.com/books/edition/The_Story_of_Philosophy/bDycoGL0Xg0C

Mill, J.S. (1863) Utilitarianism, p. 21 www.google.com/books/edition/Utilitarianism/GJc_RYBGPyEC

 TERMS TO KNOW

Absolutism
The belief that certain actions will always be morally wrong irrespective of context or consequences.

Categorical Imperative
The idea that our moral actions should be carried out as if they were universal laws.

Consequentialism
The belief that the moral value of an action or event is determined entirely by the consequences of that
event.

Deontology
The belief that moral actions are set by universal laws.

Eudaimonia
The state of achieving a good life in accordance with our purpose; flourishing.

Good Will
In Kantian philosophy, the idea that humans have an instinct toward goodness that can supersede all
other desires.

Hedonism
A theory of well-being which argues that what defines a successful life is directly correlated with the
amount of pleasure in that life.

Moral Duty
Making decisions based on our moral obligations or responsibilities.

Relativism
The belief that actions should be morally evaluated in the context and with regard to consequences.

Teleology
The belief that moral actions are determined by circumstances and outcomes.

Teleology (in Virtue Ethics)

The view that all things (including people) have a purpose and achieve goodness by fulfilling that purpose.
Aristotle’s view was that the purpose of people is to live in accordance with virtue and reason.

Utilitarianism
The belief that we can decide what is morally right or morally wrong by weighing which of our future
possible actions promotes such goodness in our lives and the lives of people more generally.

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 85
 Moral Dilemmas and Thought Experiments
by Sophia

 WHAT'S COVERED

In this tutorial, you will learn how philosophers use scenarios to consider difficult moral decisions. In
particular, you will learn about:
1. Moral Guidance
2. Moral Dilemmas
3. Determining Our Duties Through Thought Experiments

 BEFORE YOU START

In the previous tutorial, you learned three major moral frameworks:

A consequentialist approaches moral actions by considering the consequences of the decision to act
and chooses the action which causes the most benefit and the least harm
This is in contrast to a deontologist, who is not interested in consequences, but whether the rules we
use to make decisions can be universalized. An action is the right one according to a deontologist if it is
what we would recommend all people do all the time.
Finally, an adherent to virtue ethics believes right actions come from cultivating good and virtuous
people. So, a virtue ethicist will recommend the action that exhibits the virtues rather than vices.

1. Moral Guidance
Consider yourself caught in the middle of a moral dilemma. Wanting to know what to do, you may consult the
guidance offered by utilitarianism or Kantian ethics and discover that various specific actions you could
undertake are morally right or morally wrong. If you seek the advice of Aristotelian virtue ethics, you may find
cold comfort from suggestions that you act generously, patiently, and modestly whilst avoiding self-serving
flattery and envy. Rather than knowing how to live in general, you may seek knowledge of what to actually do in
this case. Virtue ethics may therefore be accused of being a theory not of helpful moral guidance, but of
unhelpful and nonspecific moral platitudes. In response, the virtue ethicist may remind us that we can learn how
to act from considering how truly virtuous people might respond in this situation, but this response raises its
own worry: how can we identify who is virtuous, or apply their actions to a potentially novel situation?

As you can see, no framework has all the moral answers, which is why philosophers still have jobs. In this
tutorial, you will consider how to approach problems from each of these three perspectives and perhaps derive
different answers to the same question.

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 86
 2. Moral Dilemmas
Kant’s theory of the categorical imperative gives us a way to “plug in” any moral question and “get out” an
answer. If we want to make an assumption explicit, such as “it is wrong to lie,” we have a basis for supporting
that as a factual statement. “Imagine a world where everybody lies all the time?”

However, in reality, we may have different categorical imperatives to consider. For example, suppose we are
high school students. The school bully accosts us in the halls and asks if we know where someone is—this
person he’s looking for is a friend of ours, and we know further that the bully wants to give him a hard time.
Here, our categorical imperative not to lie meets another categorical imperative to protect people from
violence. In this case, one categorical imperative may take value over another.

This is a moral dilemma. You may recall from Unit 3 that a dilemma is choosing between two options. The moral
dilemma is thus a situation where we have to choose between two moral principles; by doing one, we violate
the other.

Kant argued that people who were behaving irrationally—that is, not by moral duty—did not deserve our respect
and were exempt from any moral duties. In fact, he said they weren’t even human, but objects that could be
used or abused to our ends. If used in good faith, we can apply this principle to make moral decisions. We can
say, for example, that it is OK to lie to bullies because they are not living by the moral duty that one should not
hurt others, and therefore lying to them is no more immoral than lying to a street lamp.

 DID YOU KNOW

It’s historically important to acknowledge that this rationalization by Kant can lead to a moral abyss. While
on trial for war crimes, Adolf Eichmann, the man who conceived of and oversaw the Holocaust and the
orchestration of the Final Solution, claimed he was a Kantian moralist and had always tried to live by the
categorical imperative. His justification for his actions, itself grounded in Kant’s writing, was that Jewish
people weren’t rational and therefore did not deserve to be treated as humans. It is possible to misuse
moral reasoning and philosophy to justify ideas and actions that are truly horrific.

 TERM TO KNOW

Moral Dilemma
A situation in which we must ignore one moral principle to follow another.

3. Determining Our Duties Through Thought

Experiments
We have so far described what moral duty is and several theories of moral duty. However, how do we tell
whether our theory of right and wrong makes determinations that cohere with our intuition? Our moral intuition

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 87
 is our gut instinct about whether an action or belief is morally good or bad. Thought experiments can test our
theories against our intuitions. They also help us clarify and better specify our moral theories.

So, what is the philosophical device known as a “thought experiment” that we often use in theorizing about
morality? Thought experiments are hypothetical, sometimes fanciful, examples that are designed to aid our
thinking about an issue. They are sometimes unrealistic, even preposterous. In physics, for example, the famous
“Schrodinger's Cat” is a thought experiment that involves a cat in a sealed box with a flask of poison, a
radioactive element, and a Geiger counter. Even before we get to the conclusion, it is as unlikely a scenario as
you can imagine, but the conclusion is that the cat is simultaneously alive and dead, which makes it even more
preposterous. Schrodinger created the thought experiment to challenge Albert Einstein on some assumptions
about quantum physics. Fortunately, you do not have to understand it for this class!

The point is that the feasibility of thought experiments aren’t important, but they can be used to frame
discussions in many disciplines, including philosophy. They are useful as a way to think through contradictions
of moral duty.

One of the most famous moral thought experiments is the trolley car problem. There are variations on this
problem, but the basic formulation is that there is a tram (small train) on a track that is speeding and out of
control. You can see that it is headed toward several workers who are too far away to warn verbally. However,
you are standing near a switch and can divert the train to another track where a single individual is working.
That is, you must decide between harming one person on purpose or letting several people be harmed by not
doing anything. Which is your moral duty?

In a similar scenario, you are standing on a bridge that the tram will pass under before it reaches the others.
There is a heavyset man standing on the tracks. You can shove him onto the tracks and stop the train, killing
one and sparing others. (It is crucial to the problem that the man be heavy enough to stop the train, and to take
away the hypothetical solution of heroically throwing yourself in front of the trolley.) There are no right or wrong

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 88
 answers, but trying to apply the categorical imperative to these situations, and determining moral duty, can lead
to interesting discussions that help us understand real-world moral dilemmas.

 THINK ABOUT IT

What would you do in the first situation? In the second?

Most philosophers agree that we should throw the switch in the first situation and do nothing in the second,
arguing that in the second case, we would not want a universal law where it’s OK to kill other people if you
can conceive of a possibility where it would save other lives. This has led to an important distinction in
ethics between “killing” and “letting die,” made famous by ethicist Philippa Foot.
You may have entered this challenge on moral reasoning thinking, “This is easy,” because we can simply assert
opinions and there is no right or wrong answer! However, you may begin to see that consideration of moral
truths and moral duty can be as complex as deductive and inductive reasoning.

At the end of the day, we have come full circle back to Units 1 and 2. Moral reasoning and critical thinking are at
the heart of good decision making. Most arguments in the real world involve some form of normative statement
and normative conclusion. Ensuring that our arguments are valid or strong and have evidence-based premises
is crucial to arriving at a promising conclusion. Our decisions and hence our actions are guided by the
conclusions to our arguments and our reasoning about our beliefs.

Let’s return to an argument about capital punishment from Unit 2.

EXAMPLE Since the death penalty has the possibility of ending the life of innocent people, and several
people on death row in the last forty years have been found to be innocent, the state should eliminate the
death penalty.
This argument relies on both normative and declarative premises to derive a normative conclusion. It is a valid
argument against capital punishment because its premises are true and support the conclusion. Notice how the
normative premises in this argument bring us back to deontological principles of fairness. Killing innocent
people is wrong, and any law that permits it is wrong. We support these normative principles by the
deontological principles above.

In the final touchstone for the class, you will have an opportunity to construct your own normative argument like
this one that draws on elements of previous units. You will argue a moral position in standard form, using both
normative and declarative premises. You will apply the categorical imperative to show why your position is a
moral truth and not just an opinion. You will question your own assumptions to make sure your argument is not
influenced by bias. In this touchstone, as in the example above, you can see how all the components of critical
thinking we’ve covered in this course come together.

 TERM TO KNOW

Moral Intuition
Our gut feeling about whether an action is right or wrong.

 SUMMARY

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 89
 In this lesson, you learned how to refine or enrich your sense of moral guidance by considering moral
dilemmas that compel us to ignore one moral principle to follow another. Thought experiments are a
way of processing these dilemmas and considering the right moral decision.

Source: SOURCE: THIS CONTENT HAS BEEN ADAPTED FROM Ethics for A-Level.

REFERENCES

Arendt, H. (1994). Eichmann in Jerusalem : a Report on the Banality of Evil. New York: Penguin Books.

 TERMS TO KNOW

Moral Dilemma
A situation in which we must ignore one moral principle to follow another.

Moral Intuition
Our gut feeling about whether an action is right or wrong.

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 90
 Terms to Know
Absolutism
The belief that certain actions will always be morally wrong irrespective of context or
consequences.

Accidental Correlation
When there is evident correlation between two variables but no explanation of how one can
cause the other or how both can be affected by the same third factor. There are two types of
accidental correlations: nonsense and spurious.

Analogical Reasoning
The branch of inductive reasoning that discerns patterns and attempts to draw conclusions
based on comparisons with other known phenomena.

Anecdotal Evidence
Making statistical generalizations based on first- or second-hand experience.

Argument From Analogy

An inductive argument where one or more perceived similarities between two things are
presumed to indicate deeper similarities.

Background Conditions
The totality of things, many assumed, that must be in place for an event to occur.

Base Rate
The probability of the target outcome prior to the observation of any data.

Biased (Sample)
The quality of a sample not being a good/representative measure of the population, by
under-counting or over-counting some members. This may be intentional or unintentional.

Candidates
In testing for causality, the factors that may be (alone or in part) a cause for the target
outcome. These are the factors that are tracked in a log.

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 91
 Categorical Imperative
The idea that our moral actions should be carried out as if they were universal laws.

Causal Generalization
A statement about the cause of a phenomenon, that generalizes beyond the specific
observed event. They have the form: for any x, if x has the feature(s) F, then x has the feature
G.

Causal Reasoning
The branch of inductive reasoning that explores the causes of known phenomena or the
effects of known causes.

Causation
The determination that one event causes another to occur.

Concomitant Variation
The measure of two factors over time to see how one relates to the other.

Conjunction Fallacy
The reasoning error where the probability of two events happening is judged to be higher
than the probability of either event happening. This is a fallacy given that the rule for
determining the probability of two events is calculated by multiplying the probability of the
two single events together; it yields a lower probability than either single event. According to
probability theory, the conjunction should be judged to be less likely than either conjunct, but
experimental research shows that people don’t behave that way. This error can result when
the additional information makes an outcome fit better with our preconceived notions.

Connotation
An implied evaluative meaning to a word that is also descriptive.

Consequentialism
The belief that the moral value of an action or event is determined entirely by the
consequences of that event.

Conservativeness
The better explanation is the one that forces us to give up fewer established beliefs.

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 92
 Defeasibility
In inductive reasoning, the quality of an argument where new evidence might change our
conclusion.

Deontology
The belief that moral actions are set by universal laws.

Dependent
In probability, when the outcome of at least one event has an effect on at least one other
event.

Depth
Explanations should not raise more questions than they answer.

Disanalogy
A crucial difference between two things being compared, used to refute a false analogy.

Empirical Evidence
Information that can be observed (or otherwise detected by the senses).

Eudaimonia
The state of achieving a good life in accordance with our purpose; flourishing.

Evaluative Language
Words and phrases that are normative or make subjective judgments.

Explanatoriness
Explanations must explain all the observed facts.

Explanatory Power
Explanations should apply in a range of similar contexts, not just the current situation in which
the explanation is being offered.

False Analogy
An analogy that draws a broad comparison between two things that, upon scrutiny, share few
characteristics, do not have any deep resemblance, and any shared characteristic is not
relevant to the situation.

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 93
 Falsifiability
The best explanations can be tested for truth or falsehood.

Gambler’s Fallacy
The error of presuming past independent events will affect future independent events.

General (Exclusive) Disjunction Rule

P(A or B) = P(A) + P(B)

General (Inclusive) Disjunction Rule

P(A or B) = (P(A) + P(B)) - (P(A) x P(B)).

General Conjunction Rule

P(A and B) = P(A) x P(B)

Good Will
In Kantian philosophy, the idea that humans have an instinct toward goodness that can
supersede all other desires.

Hasty Generalization
Making a statistical generalization on a sample that is either inadequate in size or not
representative of the whole.

Hedonism
A theory of well-being which argues that what defines a successful life is directly correlated
with the amount of pleasure in that life.

Hidden Persuaders
The use of images, evaluative language, and other techniques to influence people without
their direct awareness.

Independent
In probability, two or more events having no effect on one another.

Inference to Best Explanation

A form of inductive reasoning where we work from a set of observed facts to come up with
the most likely explanation for an event.

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 94
 Inference to the Best Explanation
The branch of inductive reasoning that tries to explain how specific events came to occur.

Inverse
The probability of something not happening, or one minus the probability of it happening.

Jointly Sufficient
A kind of sufficient condition where there is always a target outcome when two or more
background conditions are all present.

Mind-Independent
Objective reality that does not require human thought to be true.

Modesty
Explanations should not attempt to explain anything more than the observed facts.

Moral Cognitivism
The belief that moral statements have truth value.

Moral Dilemma
A situation in which we must ignore one moral principle to follow another.

Moral Duty
Making decisions based on our moral obligations or responsibilities.

Moral Intuition
Our gut feeling about whether an action is right or wrong.

Moral Non-Cognitivism
The belief that moral statements do not have truth value, but are just feelings.

Moral Prescriptivism
The compromise between cognitivism and non-cognitivism that accepts moral statements as
essentially feelings, but we can extrapolate normative statements that do have truth value.

Mutually Exclusive

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 95
 When measuring the odds of one of multiple events occurring, this refers to events that
cannot happen simultaneously.

Necessary Condition Test

A method for identifying a necessary condition on a log tracking background conditions and
target outcomes. If the target outcome never occurs when the candidate is absent, the
candidate is likely to be a necessary condition.

Necessary Conditions (in Causality)

Factors that must be in place for an event to occur.

Negative Correlation
When measuring two factors shows that one increases as the other decreases and vice
versa.

Non-Rational
Words or actions based on emotions instead of reason.

Occam’s Razor
All other things being equal, the simplest explanation is best. The simplest explanation is the
one requiring the fewest entities and events.

Odds
The ratio of the probability of success to the probability of failures. You can calculate one
value from the other by subtracting the probability you know already from 1. Since the
possible outcome probabilities must sum to 1, the outcome you don’t already know has to be
the remaining probability.

Outcome
The possible event whose probability we are trying to determine.

Percentage
Literally “in 100,” all probabilities can be expressed as a percentage.

Population (in Statistics)

The group of things or events that are described by a generalization.

Positive Correlation

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 96
 When measuring two factors shows that both roughly increase and decrease at the same
time.

Possibilities
The total possible events, or outcomes.

Probabilistic
Statements that are true some percentage of the time, as opposed to statements that are
absolutely true or absolutely false.

Probability
A branch of mathematics concerned with describing the likelihood of future (unobserved)
events.

Quantifiable
The ability to be counted or measured.

Random Distribution
The way that random events occur over time.

Reference Class
The field of possibilities in which an event might occur, used to calculate probability.

Regression to the Mean

A statistical principle that even if there are a series of outliers or unlikely spikes and valleys in
something we’re measuring (as a consequence of random distribution), over time it will trend
toward the statistical average.

Regression to the Mean Fallacy

The error of looking for causal explanations for regression to the mean (or preceding outliers
of random distributions) that are really due to simple statistical rules.

Relativism
The belief that actions should be morally evaluated in the context and with regard to
consequences.

Representative (Sample)

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 97
 The quality of a sample being a good measure of the population, such as counting across
demographic lines in a poll.

Rhetoric
The art of persuasion.

Sample/Sampling
The portion of the population that is being directly observed or assessed to make a
generalization, or the process by which that sample is obtained.

Sampling With Replacement

In probability theory, resetting the conditions after each outcome, such as replacing each
card drawn to the deck or each numbered ball to a bin.

Sampling Without Replacement

In probability theory, having multiple outcomes without resetting the conditions, such as
drawing successive cards from a deck or numbered balls from a bin.

Simplicity
The better explanation is the one with the fewest entities and events.

Small Numbers Fallacy

The error of looking for causal explanations for extreme outliers that are really simply due to
the law of small numbers.

Statistical Generalization
A statement about how probable or likely something is based on empirical evidence.

Statistics
A branch of mathematics concerned with describing the likelihood of past (observed) events
to occur. Often it involves trying to describe the underlying causes of the observed data, and
explain any observed variation

Sufficient Condition Test

A method for identifying a sufficient condition on a log tracking background conditions and
target outcomes. If there is always an outcome when a candidate is present, it is likely to be a
sufficient condition.

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 98
 Sufficient Conditions (in Causality)
Factors that are enough by themselves for an event to occur.

Target
The event we are trying to determine the causes for in testing for causality.

Teleology
The belief that moral actions are determined by circumstances and outcomes.

Teleology (in Virtue Ethics)

The view that all things (including people) have a purpose and achieve goodness by fulfilling
that purpose. Aristotle’s view was that the purpose of people is to live in accordance with
virtue and reason.

The Law of Small Numbers

The statistical principle that small sample sizes are more likely to lead to extreme outliers on
either side of a statistical question.

Thick Term
A word that has both descriptive and evaluative meaning.

Utilitarianism
The belief that we can decide what is morally right or morally wrong by weighing which of
our future possible actions promotes such goodness in our lives and the lives of people more
generally.

Weighting
When we assign more or less probability to a particular event, either because the event is
more or less important to the likelihood of the outcome or because we’ve determined that
the event is more or less likely based on the data we’ve observed.

© 2024 SOPHIA Learning, LLC. SOPHIA is a registered trademark of SOPHIA Learning, LLC. Page 99