Logical Argument

Logic is the study of arguments. It is used to analyze an argument or a piece of reasoning,

and work out whether it is correct (valid) or not (invalid).

An argument is a conclusion with supporting statements (called premises). Logical

arguments are constructed according to certain rules so as to minimize error.

The premises and conclusions of an argument are always statements or propositions

(meanings or thoughts expressed by declarative sentences) as opposed to nonstatements
(questions, commands, or exclamations).

Statements are either true or false. Nonstatements are neither true nor false. Non
statements are never premises or conclusions.

Conclusions may be asserted (said to be true) or denied (said to be false). A conclusion is

said to be affirmed when it has been asserted based upon some argument.

The word “argument” is colloquially used to mean a disagreement, usually an unpleasant

one. These equivocal meanings frequently lead to confusion.
What may be a “good” argument in the formal sense is often a “bad” one in the
colloquial sense.
For instance, a desirable outcome in a disagreement within a family is usually NOT
establishing that one individual is right and another wrong. Rather, the desirable
outcome involves two components:
• the parties in the disagreement come to understand each other’s desires, feelings,
circumstances, etc.
• the parties in the disagreement negotiate some mutually satisfactory accommodation.
We need to distinguish between logical argument and what might be called personal
argument. They are two distinctly different processes.
• Logical argument has the purpose of providing support for statements.
• Personal argument has the purpose of changing the nature of interpersonal
relationships.
The use of one type of argument during the other may not always be helpful.
 Logical Argument - 2

Logical Argument: Inductive and Deductive Argument

There are two broad categories of argument:

• Deductive Arguments are arguments where the conclusion follows with necessity
from the premises. A deductive argument is either valid (true) or invalid (false). If
the supporting statements are true, the conclusion must be true.

Example
All students in this class are fine people.
Jamal is in this class.
Therefore
Jamal is a fine person.

• Inductive Argument involves observation of a particular sample to derive general

conclusions. Arguments in which the conclusion is derivable from the premises only
with probability are called inductive arguments. If the supporting statements are
true, the conclusion is probably true. Inductive arguments are not valid or invalid, but
we can talk about whether they are better or worse than other arguments. We can also
discuss how likely their premises are.

Example
These students are a random sample of members of that class.
All these students are fine people.
Therefore
All students in that class are fine people.

All inductive reasoning depends on the similarity of the sample and the population.
The more the similarity between the sample and the population, the more dependable
will be the inductive inference. However, if the sample is biased so as to be different
from the population, then the inductive inference will be undependable.

No inductive inference is completely accurate. Still, a good inductive inference

provides a reason to believe that the conclusion is probably true.

Formal Deductive Argument

Look at the following two arguments:

All humans have hearts. All mammals are animals.

All lawyers are human. All cats are mammals.
Therefore Therefore
All lawyers have hearts. All cats are animals.
 Logical Argument - 3

Each of these arguments is concerned about different things:

• The argument on the left has as its content Lawyers, Humans, and Hearts;
• The argument on the right has as its content Cats, Mammals, and Animals.

However, these two arguments have the same form (pattern, structure).

The pattern is

All B are C.
All A are B.
Therefore
All A are C.

This is one of many patterns (known as argument schemata) used in deductive argument.
It was identified over 2,000 years ago and is referred to by logicians as Barbara.

A classical syllogism consists of three statements (two premises and a conclusion) and
three class terms (the major, minor and middle terms). The minor and major terms must both be
in the conclusion; the middle term must appear in each of the premises; and the major and minor
terms must appear once in the premises.

In the lawyer example, the major term is “heart," the minor term is lawyer, and the middle
term is “human.”

Syllogisms

Venn diagrams allow one to make a quick test for the validity of a syllogism.

For example, we can diagram one of the Barbara arguments:

All mammals are animals.

All cats are mammals.
Therefore
All cats are animals.

We will use the letter A to represent the class term cats, B to represent mammals, and C
to represent animals.

The first premise tells us that the set of mammals is totally contained in the set of
animals, so we need to put a circle labeled B totally inside a circle labeled C.
 Logical Argument - 4

The second premise tells us that the set of cats is totally contained in the set of mammals,
so we put a circle labeled A totally inside the circle labeled B.

This demonstrates the conclusion that “All A are contained in C” or “All cats are
animals.”

The Chain Pattern of Deduction

The chain pattern of deduction involves using the conclusion of one argument as a premise for
another.

Premise 1: All B are C All depression derives from superego attacks.

Premise 2: All A are B All suicide attempts result from depression.
Conclusion 1: All A are C All suicide attempts derive from superego attacks.
Premise 3: No D are C Pre-latency children do not experience superego
attacks
Conclusion 2 : No D are A Pre-latency children do not attempt suicide.

Logical Operators

A logical operator joins two statements to form a new, more complex, statement.

The following are the logical operators:

• Conditional ( if then )
• Biconditional ( if and only if )
• Negation ( not )
• Conjunction ( and )
• Disjunction ( or )
 Logical Argument - 5

Conditional
Any two propositions, P and Q, can be joined by a conditional operator, producing the
new, complex, proposition:

If P then Q

Example: “If I’m late, then I’m in trouble” makes the statement “I’m in trouble”
conditional upon whether “I’m late.”

The proposition If P then Q is true when either P is false or Q is true. It is false only when
P is true and Q is false.

Biconditional

Any two propositions P and Q can be joined with the biconditional operator, producing
the complex, proposition:

P if and only if Q

Example: “I eat pie if and only if I bake it” means that “I eat pie” is conditional upon
“I bake pie” AND “I bake pie” is conditional upon “I eat pie.” In other
words, I only bake pie if I’m going to eat it and I only eat pie if I have
baked it.

The proposition P if and only if Q is true when both P and Q are true, or when both P and
Q are false. It is false only when one of them is true and the other false.

Negation

Any proposition P can be converted into its negative with a negation operator, producing
the proposition:

Not P

Example: “I do not like squash” is the negation of “I like squash.”

The proposition Not P is true when P is false. It is false only when P is true. The truth or
falsity of Q (or any other proposition) is irrelevant.
 Logical Argument - 6

Conjunction

Any two propositions P and Q can be conjoined, producing the proposition:

P and Q

Example: “I am late and you are late” is a combination that means both of us are late.

The proposition P and Q is true only when both P and Q are true. Otherwise, it is false.

Disjunction

Any two propositions P and Q can be disjoined, producing the proposition:

P or Q

Example: “I am late or you are late” is a combination that means that one or both of
us are late.

The proposition P or Q is true when either P or Q is true. It is false only when both P and
Q are false.

Non-syllogistic Arguments

There are formal deductive arguments that do not follow the syllogistic form (although
two of them are called syllogisms). Four frequently appearing forms are as follows:

• Disjunctive Syllogism
• Hypothetical Syllogism
• Modus Ponendo Ponens
• Modus Tollendo Tollens

Disjunctive Syllogism

Uses the Disjunction Logical Operator (or) in its first premise and the Negation Operator
(not) in its second premise. Its form is as follows:

• Either A or B Either these injuries were accidental or inflicted.

• Not A They were not accidental.
Therefore Therefore
• B They were inflicted

In the case where the first premise is not true (i.e., the first premise does not cover all
possibilities), the informal fallacy of the False Dilemma occurs.
 Logical Argument - 7

Hypothetical Syllogism

The Hypothetical Syllogism uses the Conditional Logical Operator (if. . . then) in its
premises and conclusions. It has the following form:

If A, then B If the factory closes, then more people will be unemployed.

If B, then C If more people are unemployed, then more children will be
maltreated.
Therefore Therefore
If A, then C If the factory closes, then more children will be maltreated.

Modus Ponendo Ponens

Modus Ponendo Ponens (the mood that affirms by affirming) uses the Conditional
Logical Operator (if. . . then) in its first premise.

If A, then B If Shel drinks four beers in an hour, Shel will be intoxicated.

A Shel drank four beers in the past hour.
Therefore Therefore
B Shel is intoxicated.

Modus Tollendo Tollens

Modus Tollendo Tollens (the mood that denies by denying) uses the Conditional Logical
Operator (if. . . then) in its first premise and the Negation Operator (not) in its second premise.
Its form is as follows:

If A, then B If Elwood is a good dad, he will go to the parent-teacher meeting.

Not B Elwood did not go to the parent-teacher meeting.
Therefore Therefore
Not A Elwood is not a good dad.

Fallacies

A valid argument is said to have no formal mistakes (it has the correct form). The
demonstration of the validity of an argument if it gives only true conclusions from true premises.

An invalid argument contains formal mistakes (it is incorrect form). An invalid

argument allows for a false conclusion from true premises.

Here are two arguments, their formats, and their diagrams.

• Most major child neglect involves drug abuse. Most A involves C

• Most major sexual abuse involves drug abuse. Most B involves C
Therefore Therefore
• Most major child neglect involves sexual abuse. Most A involves B
 Logical Argument - 8

It is possible for most neglect situations to involve drug abuse (A overlap C) and most
sexual abuse situations to involve drug abuse (B overlap C) without there being any association
between neglect (A) and sexual abuse (B)

• Some exhibitionists are rapists. Some A are B

• All patients on this unit are exhibitionists. All C are A
Therefore Therefore
• Some patients on this unit are rapists. Some A are C

It is possible for some exhibitionists to be rapists (A overlap B) and all patients to be

exhibitionists (C contained in A) without any patients (C) being rapists (B).

A sound argument is patterned after a valid form and has only true premises.

An argument can be unsound for three different reasons.

• It may be patterned after an invalid form or involve misapplication of a valid form

• It may contain a false premise
• It may be irrelevant or circular.
 Logical Argument - 9

Arguments patterned after an invalid schema are called Formal Fallacies.

The other two types of errors are called Informal Fallacies.

To reiterate, we can classify arguments as follows:

• Sound Arguments valid form with true premises

• Unsound Arguments
!"Formally Fallacious invalid form, misapplication of valid
form
!"Informally Fallacious valid form with false premises,
irrelevant, circular
 Logical Argument - 10

Formal Fallacies

Some commonly occurring formal fallacies (invalid or misapplied valid arguments)

include:

• Denying the Antecedent

• Affirming the Consequent
• Composition
• Division

Denying the Antecedent. This fallacy takes the form of:

• If A, then B If Moni is a registered voter, then she is over 18.

• Not A Moni is not a registered voter.
• Therefore Therefore
• Not B Moni is not over 18

The first premise is a complex conditional statement made up of an antecedent (“Moni is

a registered voter”) and a consequent (“She is over 18”).

The second premise negates (denies) the antecedent (“Moni is not a registered voter”).

The conclusion does not follow. Moni could be over 18, but never registered to vote.

This argument can be falsified by pointing out the counterexample (i.e., not all persons
over 18 are registered voters.)

Affirming the Consequent. This fallacy takes the form of:

• If A, then B If Rush is on the radio, a liar is on the radio.

• B A liar is on the radio.
Therefore Therefore
• A Rush is on the radio.

The first premise is a complex conditional statement made up of an antecedent (“Rush is

on the radio”) and a consequent (“A liar is on the radio”).

The second premise restates (affirms) the consequent (“A liar is on the radio”).

The conclusion does not follow. Another liar could be on the radio.

As previously, one may combat this argument through counter example (i.e., listing some
other liars on the radio).
 Logical Argument - 11

Composition. This fallacy takes the form of:

• A is part of M Brent, the social worker, is a member of the multidisciplinary

team
• B is part of M Bonamy, the attorney, is a member of the multidisciplinary team
• C is part of M Barbara, the pediatrician, is a member of the multidisciplinary
team
• D is part of M Boris, the child psychologist, is a member of the
multidisciplinary team
• A, B, C, and D each Brent, Bonamy, Barbara, and Boris each do their respective jobs
have quality X well.
• Therefore Therefore
• M has quality X The multidisciplinary team does its job well.
This fallacy is committed when it is argued that a property which is present or absent in
every part must be present in the whole. However, this is not necessarily the case. In the
example, each individual may be very good at his or her own job, but be unable to work with the
others

The argument also involves a semantic fallacy in that the jobs that each team member is
good at are not the same job! Similarly, the job of the multidisciplinary team is different from
each member's job and is not just the sum total of each member's job. If we had replaced the
premise "Brent, Bonamy, Barbara, and Boris are good at their respective jobs" with "Brent
performs well as a social worker; Bonamy performs well as an attorney; etc.", the fallacy might
have been more apparent. This situation is diagrammed as follows.

Each A has a quality Xa; each B has a quality Xb; each C has a quality Xc; and each D
has a quality Xd. However, Xa, Xb, Xc, or Xd do not have to relate to the aspect of A, B, C, or D
that has to do with M.

This argument is contradicted by collecting evidence as to the truth or falsity of the

conclusion. Using the multidisciplinary team example, one might point out that the members had
been unable to reach a conclusion on a single referral due to a continuing intra-team power
struggle.
 Logical Argument - 12

Division. This is sort of the inverse of the fallacy of composition,

• M has quality X The Burns-Allen couple are fun to have

around).
• A is part of M George is part of the Burns-Allen couple.
• B is part of M Gracie is part of the Burns-Allen couple
Therefore Therefore
• A and B each have quality X George is fun to have around; Gracie is fun
to have around.

This fallacy is committed when it is argued that a property which is present or absent in
the whole must be present in all parts. As with composition, this is not necessarily the case.
George may talk too much if he is alone and Gracie may be too bossy if she is alone. However,
together, Gracie restrains George’s loquacity and gives Gracie an object for her bossiness so that
they constitute a delightful couple.

As with the fallacy of composition, this argument is contradicted by collecting evidence

as to the truth or falsity of the conclusion.

Informal Fallacies

Informal Fallacies are unsound argument forms that may all be committed without
making any formal errors. Hence, they are informal fallacies. They may be organized into five
broad categories

• Appeal to Authority - These arguments take advantage of the fact that most of us are
taught to respect authority. They all involve appeals to modesty in the face of some
alleged authority.

• Irrelevant Appeals - This group of fallacies occurs when the conclusion that is
supposedly established by a set of premises is irrelevant to the point at issue.

• Confusion - This set of fallacies involves presenting an argument in such a confused or

ambiguous fashion that it is not clear what it means or what it is supposed to prove.

• Faulty Classification - This set of fallacies all involve errors associated with the
formation of classes, sets, groups, or other aggregations.
• Begging the Question - This set of informal fallacies all involve assuming the truth of a
question rather than proving it.
 Logical Argument - 13

Appeals to Authority (Pseudoauthority)

The traditional Latin name for this class of fallacies is argumentum ad verecundam. This
literally means “argument to modesty.” They take advantage of the fact that most of us are
taught very early in life to yield to authority. Some common examples of the fallacy of
Pseudoauthority include:

Irrelevant Authority. When a reputable authority in one area is presented as an authority

in an entirely different area where he or she has no special expertise.

For example:

Nobel Prize winner William Shockley has said that intelligence is an hereditary
characteristic. Therefore, it has been argued that any racial differences in performance on
intelligence tests must be due to genetic deficiencies.

The problem with this argument is that expertise in one area (semiconductor research)
does not imply expertise in another (assessment of cognitive functioning, genetics). In the case of
Shockley, his opinion on the nature of "intelligence" carries no more weight than any other non-
expert's opinion.
Aphorism. When one uses some aphorism, cliché, maxim or proverb in the absence of a
legitimate argument. This is basically a way of avoiding seeking actual evidence in favor of a
prejudice "supported" by an aphorism that may not always be true.
For example, I may oppose setting up a committee to seek community input by saying
"An elephant is a mouse designed by a committee." I imply that a committee would be
inefficient without providing any evidence that such is the case. This is best refuted by evidence
to the contrary. In this case, you could point out that one of the masterpieces of English literature
was created by a committee - the King James translation of the Bible.

Apriorism. An a priori fallacy occurs when it is argued prior to any investigation that
certain events must or must not occur according to some theory or point of view.

For example:
Psychoanalytic theorists argued that clinical depression was not possible for pre-
adolescent children. This was because psychoanalytic theory required that clinical
depression result from a failure in correspondence between the real self and the ideal self.
Since a stable self-representation was not supposed to develop until adolescence, children
could not demonstrate the full clinical syndrome of depression.
A statement does not have to be true simply because a theory says it is. In fact, pre-
adolescent children have been found to experience the full clinical syndrome of depression.
 Logical Argument - 14

Irrelevant Appeals

This class of fallacies was identified over two thousand years ago by Aristotle. The
traditional name for then is ignoratio elenchi (literally, “ignorant refutations”). These fallacies
all involve the use of premises that are irrelevant to the conclusion. The following are a small
sample of irrelevant appeals:

Appeal to Pity. The Appeal to Pity (argumentum ad misericordium) is committed when

one tries to persuade someone to a point of view by arousing feelings of pity.

For example:
"If I get a C in this course, I'll be thrown out of the program." The implied conclusion is
that I should be given a grade higher than C.
Before we go further, I'd like to note that an appeal to pity is not always irrelevant.
However, the burden rests upon the individual making the argument to demonstrate relevance.
Specifically, he or she must show why compassion is the appropriate criterion for making the
particular decision

In this instance, the burden was not met. The issue was not the bad consequences to the
student, but the fact that the student engaged in the behavior that brought on those consequences
and is now trying to evade responsibility for his or her own acts.

Further, compassion is a two-edged sword. If we ought to have compassion for the

student, we similarly ought to have compassion for his or her future clients who would be
receiving services from an inadequately trained social worker

Appeal to Ignorance. (argumentum ad ignorantiam - This is an argument that

illegitimately attempts to shift the burden of proof from the person making the argument to the
person hearing the argument. Basically, it is argued that the absence of evidence against a point
of view must be counted as evidence for it

Elwood said, "All homosexuals were originally recruited by child molesters."

Jake asked, "Really? I find that very hard to believe.”
Elwood replied, "Can you prove that I'm wrong?"
Elwood is attempting to place Jake on the defensive. The appeal to ignorance is, in fact, a
form of intimidation. Elwood is trying to force Jake to the conclusion that a same sex orientation
is the result of child molestation simply because Jake cannot immediately disprove the assertion.
However, in logical argument, the burden of proof is on the individual making the
assertion, not the listener.
 Logical Argument - 15

Abusing the Person. (argumentum ad hominem)- This fallacy involves shifting the
attention from the issue and onto the person who is making the argument.

These personal attacks come in many forms. Abuse is the simplest. Others include bad
seed (the descendants of a "bad" person are not to be trusted), bad connections (people who
associate with "bad" people are not to be trusted), genetic fallacy (if a person's motives for
supporting a position are "improper", then the position is a bad one) and poisoning the well
(making an accusation where replying to the accusation serves to create an illusion that the
accusation is true).

Here are two examples:

• “You talk too much.”

This is poisoning the well. If you argue that you do not talk too much, you are talking
and have proven the point. If you hold silent, you have conceded the point that you talk
too much.
• “The reason why you disagree with your boss is that she is a woman and your arguments
with her are derived from your turbulent relationship with your mother.
This is the genetic fallacy. The speaker attempts to prove the argument false by attacking
its source – its genesis. The intrusion of psychoanalytic formulations into popular
culture has tended to make the genetic fallacy popular. While it may be true that a
person’s motives can weaken the person’s credibility, they are still irrelevant to the
credibility of the argument. Whether or not you get along with your mom, the boss can
still be wrong.
Appeals to Confusion

This set of fallacies involves presenting an argument in such a confused or ambiguous

fashion that it is not clear what the argument means or what it is supposed to prove. Some
frequently occurring appeals to confusion are:

Equivocation. This fallacy occurs when the double meaning of a term is played upon in a
misleading fashion. The perpetrator of this fallacy shifts the meaning of a key word over the
course of an argument.

For example, " The editors of this newspaper have a duty to publish such news as it is in
the public interest to have published. There can be no doubt about the public interest in the
identity of the woman who alleged that she was sexually assaulted at the estate of a wealthy oil
executive. Therefore, the editors of this paper would have failed in their duty if they had
refrained from publishing the woman's name."
The twist here is in the meaning of the term "public interest." "Public interest" means
"public welfare" in the first sentence, while it means "what many people are interested in" in the
second sentence.
 Logical Argument - 16

Exceptions To The Rule. This fallacy is based upon the aphorism "The exception proves
the rule." It is committed when an individual claims that some rule or generalization has been
established because of the existence of an exception or exceptions. The aphorism is fatuous. An
exception does not prove a rule. It disproves a rule.

For example,
Elwood: “Republicans are all narrow-minded, hardhearted jerks.”
Edith Anne: “My brother-in-law is a Republican and he is generous and broad-
minded.”
Elwood: “Well, he’s the exception that proves the rule.”
No, he is the exception that disproves the rule. All Republicans are not narrow-minded,
hardhearted jerks. Of course, whether most Republicans are narrow-minded, hardhearted jerks is
still open for discussion.

Answering Questions With Questions. When a legitimate question is answered with

another question, the fallacy of confusing by answering a question with a question has been
committed.

For example,
Michael: "What data supports the notion that your program is actually helping
clients?"
George: " What makes you think that everything can be reduced to grids and
graphs? Don't you recognize that our clients are unique human beings with
unique problems?"
The fallacy of answering a question with a question is similar to shifting the burden of
proof by Appeal to Ignorance. However, while the Appeal to Ignorance involves answering the
question "What evidence is there for your position?" with "What evidence is there against it?",
answering questions with questions involves a refusal to address the question at all and an
attempt to shift the discussion away from the issue raised by the question.

Often this means that the individual has some very good reasons for not wanting to
answer the question.

Faulty Classification

This set of fallacies all involve errors associated with the formation of classes, sets,
groups, or other aggregations. In most cases they involve looking at irrelevant similarities or
differences and developing conclusions from them. All of these fallacies, to some degree,
involve oversimplification.

Some treatments of fallacies include Composition and Division in this category. They are
appropriately placed here, but also can be placed in category of formal fallacies which is what I
have done.
 Logical Argument - 17

Continuum. This fallacy is committed when it is argued that, because of a continuum of

differences between two extremes, there is no "real" difference.

For example:
Since there is no abrupt and obvious boundary between appropriate and abusive child
discipline behaviors, one might argue that there are no "real" differences between the
extremes of appropriate and abusive discipline.
This fallacy relies upon attending to one characteristic while ignoring the importance of
the magnitude of the characteristic. One may refute such arguments by pointing out the obvious
differences in the extremes (e.g., there is a difference between a child who is beaten to death and
a child who receives one swat on the bottom for running out in the street). One ought also to
concede that there is a "gray area". If we do not, we run the risk of committing a False Dilemma
fallacy (discussed at the bottom of this page).

Golden Mean. This fallacy is committed when it is argued that the mean or middle view
between two extremes must be right simply because it is the mean or middle view. This plays
upon a cultural bias that tends to view “extremism” as being somehow dangerous.
Of course, it is sometimes appropriate to choose the middle ground. It is more conducive
to survival to eat moderately than to starve on the one hand or gorge oneself on the other.
However, this is not because there is some inherent virtue associated with the middle. Whether
the middle is appropriate depends upon the circumstances. Obviously the middle position is not
appropriate when we drive an automobile or we would drive down the middle of the street rather
than drive down the right side to avoid extremism.

Here's another example:

Moe: Gays and Lesbians are dangerous and should not be permitted to work in
occupations that bring them in contact with children.
Larry: Gays and Lesbians are no more dangerous than straight folks. Our sexual
orientation should be no bar to working with children.
Curly: Let's not be extremist about this. The middle way is the best way. Here is
what we will do. Gays and Lesbians will be able to work with children
under close supervision. We'll let parents know which staff are gay or
Lesbian so that they may choose whether to have their children be in their
classes.
The "middle position" is still extremely problematic. Since there is no evidence of danger
to children being associated with sexual orientation of teachers, the middle position still involves
unnecessary and oppressive invasion of privacy.

False Dilemma. The False Dilemma fallacy is committed when it is argued that only one
of a restricted number of arguments is true where they do not represent all legitimate arguments
and/or the arguments are not mutually exclusive. The argument may be presented as a disjunctive
syllogism:
 Logical Argument - 18

“Moni is depressed. There are basically two choices: She can undergo long-term
psychotherapy or she can commit suicide. I think we can agree that psychotherapy is the
preferred alternative.”
In the first place, most instances of depression are not associated with suicide attempts, so
the disjunction (long-term osychotherapy or suicide) is not true. In the second, long-term
psychotherapy is not the only appropriate response to a depression problem. There are other
alternatives, many of which are superior to long-term psychotherapy.
Begging the Question

This set of informal fallacies involves assuming the truth of a question rather than
proving it. The traditional name is petitio principii .

Alleged Certainty. This fallacy occurs when a claim in question is qualified by a phrase
(e.g., "everybody knows") that tends to persuade without proving that the claim is beyond doubt.

For example:
"Everybody knows that if you simply treat the symptom without treating the underlying
disorder of psychic structure, you will just get symptom substitution."
Still, even if "everybody knows", it does not follow that what "everybody knows" is true.
In fact, Freud reported symptom substitution only in cases of hypnotherapy. The evidence from
the empirical clinical practice literature is that symptom substitution is a comparatively rare
phenomenon.
This fallacy is also the basis for much bigoted belief. It is best dealt with by examining
and presenting the evidence. Probably the best response to "everybody knows…" and "You have
to admit that…" are statements like "I don't know any such thing" and "I have to admit nothing of
the kind." They focus attention on the evidence or lack thereof for the statement.
Question Begging Definitions. When one uses question-begging definitions, terms of a
presumably factual, but questionable assertion are defined in such a way that the assertion cannot
fail to be true.

For Example:
Phyllis: All Christians believe that abortion is murder.
Philippa: I am a Christian and I don't believe that abortion is murder. In fact, I
believe it is wrong to force a woman to carry an unwanted fetus.
Phyllis: You are not a Christian because all real Christians believe that abortion is
murder.
Phyllis has defined "real" Christianity such that opposition to abortion is a defining
feature of "Christianity." However, Phyllis has not supported the statement that "real" Christians
(as contrasted with "unreal" Christians) oppose abortions beyond saying it is so. Since this is the
conclusion to be proven, the argument has failed
 Logical Argument - 19

Circular Reasoning. This fallacy involves using the conclusion you are trying to reach
as a premise or presupposing the conclusion in a premise.

For example:
" Unless we have an internal representation of a loved one (an object), then we cannot
love that person. But, we do love. Therefore, these internal representations (objects)
exist."
The first sentence of this argument contains an unproven assertion which is also the
conclusion. Namely, it asserts that an internal representation of an individual is a prerequisite to
loving that individual. However, the question at issue is whether such representations exist. The
argument starts with the assumption that they do exist and proceeds on this basis to prove that
they exist. Thus, we call them circular arguments.
 Logical Argument - 20

Inductive Argument

Inductive arguments may be thought of as falling into two categories. The first category
involves those arguments that are based in the assumption that the universe is consistent. This is,
of course, one of the central assumptions of the Scientific Method.

Inductive arguments that do not assume consistency have premises that use statistical
reasons to support conclusions.

We shall call arguments that assume that the universe has some consistency consistency
arguments and those that do not require this assumption statistical arguments.

To make the distinction between statistical and consistency arguments clearer, let us
examine some examples of each.

Here is a consistency argument.

None of the 100 social workers we investigated has behaved unethically.
Therefore,
If we investigate another social worker, he or she will not have behaved
unethically.
This argument is a form of induction by enumeration. Even though the first premise
involves the collection of statistics, the argument is not a statistical argument. The reason is that
the conclusion refers to an individual who was not one of the group of social workers in the
premise. Thus, we must assume that the incidence of unethical behavior in the social workers not
investigated is consistent with the incidence in the group surveyed. In other words, we must
assume that the universe is a consistent place to be able to reach our conclusion. This places
simple induction within the category of consistency arguments.

Here is a statistical argument.

Very few social workers in Ingham County behave unethically.
Jake is a social worker in Ingham County.
Therefore,
Jake does not behave unethically.
This is a statistical syllogism. The conclusion refers to an individual who is a member of
the group discussed in the first premise. We do not need to make any additional assumptions
(such as consistency) to conclude that it is unlikely that Jake behaves unethically.

We shall look at four types of inductive argument:

Statistical Syllogism

Statistical Generalization

Induction by Enumeration

Induction by Analogy
 Logical Argument - 21

Statistical Syllogisms

The statistical syllogism involves making an inference about a member of a group using
statistics about the group of individuals. For example, here is a statistical syllogism that belongs
to the category of statistical inductive arguments.
99.44% of social workers employed by the state complain about filling out forms.
Elwood is a social worker employed by the state.
Therefore,
Elwood complains about filling out forms.
On statistical grounds, the conclusion is likely given the premises. In fact, we can
accurately give the inductive probability as 99.44% without any recourse to an assumption of
consistency.

The standard format for the statistical syllogism is -

p of A is B.
x is A.
Therefore,
x is B.
where p is a proportion and x is an element of A. The statistical syllogism usually takes
the preceding form only when p > 0.5. When p < 0.5 it takes the following form -
p of A is B.
x is A.
Therefore,
x is not B.
When p = 1, an inductive syllogism becomes deductive.
All of A is B.
x is A.
Therefore,
x is B.
The premises in a statistical syllogism do not need to be numerically precise.
Most social workers employed by the state do not have an M.S.W.
Jake is a social worker employed by the state.
Therefore,
Jake does not have an M.S.W.
Since we are not making inferences about individuals outside the group in the first
premise, this is also a statistical inductive argument. While we do not know the exact inductive
probability, we know that it is greater than 0.5 and less than 1.0.
 Logical Argument - 22

Statistical Generalization

Statistical generalization is a process distinct from the use of statistical syllogism. A

statistical syllogism involves an inference from statistics about a group of individuals to a
member of the group. Statistical generalization involves randomly selecting a sample of
individuals from a larger group, collecting statistics on this smaller sample, and using these
statistics to make inferences about the larger group.

Even though conclusions are being drawn about entities not included in the group being
observed, this is still statistical induction. The reason is contained in the procedure for choosing
the group to be observed. Since we are using a random sample, we can determine the inductive
probability of our conclusion using mathematical procedures. We do not have to assume any
natural consistency. Therefore, statistical generalization is a form of statistical rather than
consistency induction.

Here is a statistical generalization with a fairly high inductive probability.

About 26% of 1,000 social workers randomly selected from Michigan social
service agencies are male.
Therefore,
Between 23% and 29% of social workers in Michigan social service agencies are
male.
There are mathematical procedures for computing the inductive probability if a random
sample has been used. In this instance, the inductive probability is about 95%.

Here is the general form of statistical generalization -

p of n randomly selected A is B.
Therefore,
About p of all A is B.
The number n is the size of the sample. A is the group (population) from which the
sample was drawn. p is the proportion of the sample that had characteristic B.

The validity of inferences made by statistical generalization is necessarily dependent

upon the use of random sampling techniques. The effect of type of sample on conclusion validity
will be extensively discussed in subsequent units of this course.

If the sample is random, two characteristics will affect inductive probability: the sample
size (n) and the precision of the conclusion.

If we increase n, this strengthens the argument's premise in a relevant way which

increases the argument's inductive probability.

If we weaken the conclusion, we will increase the argument's inductive probability.

Notice that the conclusion in the example was "Between 23% and 29% of social workers in
Michigan social service agencies are male". This is a weaker conclusion than "Exactly 26% of
social workers in Michigan social service agencies are male".
 Logical Argument - 23

It is very unlikely that a random sample would contain exactly the same proportion of Bs
(males) as the population from which it was drawn. So, if we want our argument to have higher
inductive probability, we must allow a margin of error in our conclusion. This is what the range
of 23% to 29% signified. In mathematical statistics, this is called a confidence interval.

Induction by Enumeration

Another name for this category of induction is inductive generalization. An example of

this type of argument is
About 76% of the 54 clients who completed our program last year did so
successfully.
Therefore,
About 76% of the clients who complete our program will do so successfully.
The form of this argument is
p of n observed A are B.
Therefore,
About p of all A are B.
The number n is the number of elements of A that have been observed. p is the
proportion of n that had characteristic B. In the example, p is 75%, n is 54, A is clients observed,
and B is "successfully completed our program".

Induction by enumeration differs from statistical generalization in that it does not use a
random sample in its premise. Since this is the case, induction by enumeration is a consistency
argument. It depends upon the unstated assumption that the universe will be consistent with what
has already been observed. In the case of the example, the assumption is that the overall
successful completion rate will be the same as the rate for last year.

All else being equal, induction by enumeration arguments will have a lower inductive
probability than statistical generalization arguments. The induction by enumeration arguments
have the additional liability of not permitting the computation of an exact probability (which we
can do for statistical generalization).

Some authorities argue that we can increase the inductive probability of induction by
enumeration through increasing the number of observations. This will be the case if there is no
systematic bias in our observations. However, if there is bias, we may give the appearance of
higher inductive probability to an argument that has a false conclusion.

For example, consider this argument -

78% of 50 Michiganians surveyed said they supported school vouchers.
Therefore,
About 78% of all of all Michiganians say they support school vouchers.
 Logical Argument - 24

Now, consider this argument -

78% of 500 Michiganians surveyed said they supported school vouchers.
Therefore,
About 78% of all of all Michiganians say they support school vouchers.
The second argument appears to have a higher probability than the first since it is based
upon a larger sample. However, it may be less true than the first if the 500 individuals were all
attendees at a conference for private and parochial school administrators. So, more observations
does not necessarily imply greater accuracy.

Another form of the enumeration argument is

All of n observed A are B.
Therefore,
All A are B.
This is the way that scientific laws are justified. Such laws fall into the category of
contingent knowledge and require only one instance where an A is not B to invalidate them.

Another form of induction by enumeration is sometimes called simple induction:

p of n observed A are B.
Therefore,
If we observe one more A, it will be a B.
99% of 100 teenagers observed at the fine arts magnet school wear black.
Therefore,
If we observe one more teenager at the fine arts magnet school, he or she will be
wearing black.
However, if we strengthen the conclusion statement (as in the following argument), the
inductive probability of the argument becomes lower:
99% of 100 teenagers observed at the fine arts magnet school wear black.
Therefore,
99% of teenagers at the fine arts magnet school wear black.
Induction by Analogy

An example of this type of argument is

Pamina is a yoga instructor, a member of People for the Ethical Treatment of
Animals, and a member of the Hogtown Granary Food Co-op.
Papagena is a yoga instructor, a member of People for the Ethical Treatment of
Animals, and a member of the Hogtown Granary Food Co-op.
Papagena is a vegetarian.
Therefore,
Pamina is a vegetarian.
 Logical Argument - 25

In such an argument we observe that entity x has many characteristics C1, C2, C3, . . Cn in
common with entity y. We further observe that y has characteristic D. Since x and y are analogous
in so many other ways, we conclude that x has characteristic D as well. The form of this argument
is
C1x & C2x & C3x.
C1y & C2y & C3y.
Dy.
Therefore,
Dx.
Where
C1 = yoga instructor,
C2 = member of People for the Ethical Treatment of Animals,
C3 = member of the Hogtown Granary Food Co-op,
x = Pamina,
y = Papagena, and
D = vegetarian.
Since there is no statistical basis produced to connect the similarities in characteristics,
the induction by analogy argument depends upon an assumption of consistency.

The inductive probability of analogy arguments depends upon the relevance of properties
C1, C2, C3 to property D. For the example argument, the properties are probably related to each
other, so the inductive probability is fairly high.

However, consider this argument:

Papagena listens to National Public Radio, likes Vernor's ginger ale, and
watches hockey.
Papageno listens to National Public Radio, likes Vernor's ginger ale, and
watches hockey.
Papageno likes mystery novels.
Therefore,
Papagena likes mystery novels.
This is a little more concerning. It is not clear that listening to NPR, liking Vernor's, and
watching hockey are relevant either separately or together to liking mystery novels. The premises
of an argument must (among other things) be relevant to the conclusion if the conclusion is to be
supported. Lack of demonstrated relevance will result in an argument with low inductive
probability.

Induction by analogy collapses when two similar objects are different in a way that
affects whether they both share the property under consideration. This leads to an inductive
fallacy (discussed later) called Questionable Analogy. It might also be thought of as a form of the
Exclusion fallacy in cases where information about the difference is available, but withheld.
 Logical Argument - 26

Inductive Fallacies

All inductive reasoning depends upon one of two assumptions:

• Consistency: meaning that there is a consistency inherent in the situation under

discussion that permits generalization from observed events to unobserved events.

• Statistical Validity: meaning that the samples discussed in the premises have been
randomly drawn thus allowing for exact evaluation of the inductive probability of the
argument.

One or the other of these assumptions are necessary for the development of an argument
with high inductive probability. As with deductive arguments, the argument must also be in valid
form, and the premises must be true and relevant to the conclusion.

Some common inductive fallacies include:

Fallacy of Exclusion

Questionable Analogy

Hasty Generalization

Unrepresentative Sample

Inductive inferences are probabilistic and, hence, not perfect. This means that even a well
supported inductive inference may occasionally be false. Even when the premises are true, the
conclusion might be false. This is one of the reasons why we tend to discuss the probability of
inductive arguments rather than their validity.

Exclusion

This fallacy was mentioned when we discussed induction by enumeration. This was
where the conclusion was that about 78% of all of all Michiganians say they support school
vouchers. This was based upon a survey of 500 Michiganians. The information that was excluded
was that the 500 individuals were all attendees at a conference for private and parochial school.

This fallacy occurs when information that would affect a conclusion is left out of the
argument. The fallacious argument appears sound unless one knows that the information has
been excluded.

Here is another example:

Tim will probably receive an A in this course because he has received A's in
every course he has taken so far.
 Logical Argument - 27

This is a form of the induction by enumeration argument:

Tim has received A's in every course he has taken so far.
Therefore,
If he takes one more course, he will probably receive an A in that course.
All of n observed A are B.
Therefore,
If we observe one more A, it will be a B.
As it stands, this argument has pretty good inductive probability. The premise (Tim has
received A's in every course he has taken so far.) is strong; the conclusion (If he takes
one more course, he will probably receive an A in that course.) is fairly weak; and the
premise is highly relevant to the conclusion.

The exclusion fallacy works by reducing the relevance of the premise to the conclusion,
thereby reducing the inductive probability of the argument.

For example, let us imagine that some information has been excluded from this argument.
Specifically, we have not been told that all the courses previously taken by Tim were "easy"
courses, and the next course will be a "difficult" one. This excluded information would weaken
the relevance of the premise to the conclusion and reduce the probability of the argument.
Tim has received A's in every "easy" course he has taken so far.
Therefore,
If he takes a "difficult" course, he will probably receive an A in that course.
This argument is falsified by producing the missing evidence and demonstrating that it
changes the inductive probability of the argument. It is not sufficient just to assert that there
might be excluded information. The information must be produced. Having done so, simply
showing that some evidence was excluded is also insufficient. One must be show that the
missing evidence affects the inductive probability of the argument.

Questionable Analogy

This fallacy involves a misuse of the induction by analogy argument. When we use
analogical reasoning, we infer from an observed similarity of two objects on a set of observed
properties that they are similar on another property that was not observed. However, this
inference requires that the observed properties must be relevant to the unobserved property.

Here is the questionable analogy we looked at earlier.

Papagena listens to National Public Radio, likes Vernor's ginger ale, and
watches hockey.
Papageno listens to National Public Radio, likes Vernor's ginger ale, and
watches hockey.
Papageno likes mystery novels.
Therefore,
Papagena likes mystery novels.
 Logical Argument - 28

It is not clear that the observed properties

• listening to NPR,
• liking Vernor's, and
• watching hockey
are relevant to liking mystery novels.
An analogy fails when two similar objects are different in a way that affects whether they
both share the property under consideration.
Here is another example:

Government and business carry out many of the same tasks. So just as business
must be sensitive primarily to the bottom line, government must do likewise.

This argument could be stated with greater specificity as follows:

Government receives money (taxes) from individuals and provides services.
Businesses receive money from individuals and provide services.
Businesses must be primarily sensitive to the bottom line (profit and cost).
Therefore,
Government must be primarily sensitive to the bottom line (profit and cost).

However, the objectives of government and business are different. The primary objective
of a business is to show a profit. The primary objective of a government is to provide services to
its citizens. Since the objectives are different, it is probable that they will not give the same
weight to the bottom line.

(This example is derived from one found on Stephen Downes' on-line Fallacy Page.)

As mentioned previously, I believe that someone could make a good case that this is a
special case of the exclusion fallacy applied to analogical reasoning. At any rate, this fallacy is
uncovered by showing that the two objects under comparison are different in a way that will
affect the conclusion of the argument.

Hasty Generalization

This fallacy involves misuse of the statistical generalization and induction by

enumeration arguments. It is sometimes called the fallacy of insufficient statistics. This fallacy
occurs when one reaches a conclusion before enough data have been collected to justify the
conclusion. In other words, the sample size is too small to support the conclusion.

The following is a statistical generalization argument:

About 40% of 10 social workers randomly selected from Michigan social service
agencies are male.
Therefore,
Between 30% and 50% social workers in Michigan social service agencies are
male.
 Logical Argument - 29

Compare this argument with the example given for statistical generalization.

There are some features of the new example that would appear to make it a more likely
argument than the old one. The conclusion of the new argument (30% - 50%) is weaker than the
conclusion of the old one (23% - 29%), which would raise the inductive probability of the new
argument.

The sample proportion in the new argument (40%) is higher than the sample proportion in
the old argument (26%), which would tend to strengthen the premise and raise the inductive
probability of the argument.

However, the sample size in the new argument (10) is much, much lower than the sample
size in the old argument (1000). This substantially weakens the premise in the new argument and
lowers the inductive probability of the new argument. The old conclusion ("Between 23% and
29% of social workers in Michigan social service agencies are male.") had an inductive
probability of about 95%. Because of the small sample size, we get an inductive probability of
less than 50% for the conclusion of the new argument ("Between 30% and 50% of social workers
in Michigan social service agencies are male.").

Here is an induction by enumeration example:

Two of the last three shoplifters referred were truant schoolchildren.
Fred was just referred for shoplifting.
It is very likely that Fred is a truant as well.

With a sample size of three, our inductive probability is going to be pretty low. It is not
unreasonable for a sample this small to be atypical, especially when it is a non-random sample.
This is the problem with anecdotal illustrations. They may be atypical.

This argument may be contradicted by noting that small samples are subject to large error
when making inferences about population characteristics. This can be demonstrated
mathematically by noting the effect of sample size on the standard error of the mean.

Unrepresentative Sample

This fallacy also involves misuse of the statistical generalization and induction by
enumeration arguments. It is also called the fallacy of biased statistics. It occurs when one draws
a conclusion based upon a sample that is known to be biased, or for which there is good reason to
believe is biased. This essentially involves making inferences to a particular population from a
sample drawn from a different population.

Here is the complete version of the unrepresentative sample fallacy we looked at earlier:
78% of 500 Michigan private and parochial school administrators surveyed said
they supported school vouchers.
Therefore,
About 78% of all of all Michiganians say they support school vouchers.
 Logical Argument - 30

The bias inherent in this argument is fairly obvious. The sample has been drawn from one
Michigan subpopulation (private and parochial school administrators) to make inferences about
the total population of Michigan. It is likely that the proportion of private and parochial school
administrators favoring vouchers would be higher than the proportion found in the general
population.

Another example of biased statistics was the notorious Literary Digest poll that predicted
that Alf Landon would defeat Franklin D. Roosevelt in the 1936 presidential election. Roosevelt
won by a landslide.

The sample for the survey was drawn from telephone directories and automobile
registrations. During the Depression of the 1930s, many people did not have an automobile or a
telephone. The sample was thus biased in favor of those better off individuals who had a
telephone and/or automobile. These individuals were more likely to vote Republican than were
less well off individuals.

This argument is falsified by showing how the sample is different from the population
and demonstrating that it changes the probability of the conclusion. It is NOT sufficient simply
to show that the sample is in some way different. One must be show that the difference is
relevant.