INDUCTIVE & DEDUCTIVE LOGIC

Complete Exam Notes

Simple Language · Full Examples · Comparisons · Practice Questions

📚 Topics Covered in These Notes:

TOPIC 1: Inductive vs. Deductive Logic — History, Differences, and Importance
TOPIC 2: The Three Kinds of Inductive Arguments
→ Generalisation | Argument from Analogy | Causal Inference
TOPIC 3: Topics of Inductive Reasoning
→ Random Variation | Margin of Error | Confidence Level | Fallacies in
Inductive Reasoning
 TOPIC 1: INDUCTIVE VS. DEDUCTIVE LOGIC

Background and History

Logic is the foundation of all rational thinking. It is what allows humans to reason clearly, make decisions,
build arguments, and reach conclusions. There are two main types of logical reasoning: deductive logic
and inductive logic. Both have been used for thousands of years, and understanding both is essential for
critical thinking.
Deductive logic was developed by the ancient Greek philosopher Aristotle, who created a system called
syllogistic reasoning — where you link general rules together to reach specific conclusions. For example:
'All men are mortal. Socrates is a man. Therefore, Socrates is mortal.' This was the dominant form of logic
for over two thousand years.
Inductive logic became more prominent with the English philosopher Francis Bacon in the 1600s. Bacon
argued that instead of starting with general rules and deducing specifics from them, scientists should start
with specific observations from the real world and use them to build general conclusions. This is the
foundation of the scientific method.

Feature Deductive Logic (Aristotle) Inductive Logic (Francis

Bacon)
When developed Ancient Greece (~350 BC) 1600s —
Renaissance/Scientific
Revolution
Starting point General rules / principles Specific real-world observations
Direction General → Specific Specific → General
Key figure Aristotle Francis Bacon
Main use Mathematics, formal proofs, Science, research, everyday
legal deduction reasoning

What is Deductive Logic?

📌 Deductive Reasoning
Deductive reasoning starts from a general rule that is already known to be true, applies it to a
specific situation, and reaches a conclusion that is 100% certain — as long as both premises
are true and the logic is correct. If the premises are true and the logic is valid, the conclusion
CANNOT be false. It is guaranteed.

Think of deductive reasoning like a mathematical equation. If you know that 2 + 2 = 4, and someone gives
you 2 apples and then 2 more apples, you can deduce with complete certainty that you now have 4 apples.
There is no possibility of being wrong — as long as the starting facts are true.

💡 Classic Example
P1 (Premise 1): All humans are mortal. ← General rule P2 (Premise 2): Socrates is a
human. ← Specific case Conclusion: Therefore, Socrates is mortal. ← Certain
conclusion If both premises are true, the conclusion MUST be true. No exceptions.
 💡 Another Example
P1: There is no mail delivery on holidays. P2: Tomorrow is a public holiday. Conclusion:
Therefore, there will be no mail delivery tomorrow. This is deductive. The conclusion is
guaranteed if both premises are true.

Valid vs. Sound Deductive Arguments

Two very important terms in deductive logic are VALID and SOUND. These are often confused, so pay
close attention to the difference.

Term What It Means Example

VALID The logical structure is correct P1: All birds can fly. P2: A
— the conclusion follows from penguin is a bird. Conclusion: A
the premises. This is true penguin can fly. — The
EVEN IF the premises are structure is valid, even though
false. P1 is false.
SOUND Valid AND the premises are P1: All mammals breathe air.
actually true. A sound argument P2: A whale is a mammal.
gives a guaranteed true Conclusion: A whale breathes
conclusion. air. — Valid structure + true
premises = sound.
Invalid The conclusion does NOT P1: Some birds can fly. P2: A
follow logically from the bat is not a bird. Conclusion:
premises, regardless of Bats can't fly. — The logic is
whether the premises are true. broken.

⚠️ Key Distinction to Remember

Valid = Correct logic structure (but premises could be false) Sound = Valid + All premises are
actually true You can have a valid argument that is NOT sound (if a premise is false). But you
CANNOT have a sound argument that is NOT valid. Sound is the gold standard — it gives you
a conclusion you can truly rely on.

What is Inductive Logic?

📌 Inductive Reasoning
Inductive reasoning starts from specific observations or examples, notices a pattern, and
draws a general conclusion. The conclusion is probable — it is likely to be true, but it is
NEVER 100% guaranteed, even if all the premises are true. New evidence could always come
along that contradicts the conclusion.

Think of inductive reasoning like a detective building a case. A detective doesn't have direct proof — they
gather clues, notice patterns, and conclude what probably happened. The conclusion is their best, most
reasonable explanation — but new evidence could change it.

💡 Classic Example
Observation 1: Every swan I have seen in Europe is white. Observation 2: Every swan my
friends have seen is white. Observation 3: Every swan in every European painting is white.
 Conclusion: Therefore, all swans are white. For centuries, Europeans believed this — until
they discovered BLACK swans in Australia! This shows perfectly why inductive conclusions
are probable, NOT certain.

💡 Science Example
Doctors observed tens of thousands of patients over decades. They noticed that heavy
smokers developed lung cancer at far higher rates than non-smokers. From this pattern of
specific observations, they drew the general conclusion: Smoking causes lung cancer. This is
inductive — built from many real-world observations. Extremely well-supported, but technically
a probable conclusion, not a mathematical certainty.

Strong vs. Weak Inductive Arguments

Since inductive arguments can never be 100% certain, we judge them by how STRONG or WEAK they
are. A strong argument has a lot of good evidence. A weak argument has very little evidence.

Type What It Means Example

STRONG Large, diverse, representative 'In a study of 50,000 patients
sample. The evidence makes across 30 countries, 92% who
the conclusion very likely. took this vaccine did not
develop the disease.' — Very
strong evidence.
WEAK Small, biased, or 'My two friends tried this
unrepresentative sample. The vaccine and it didn't work for
evidence barely supports the them, so vaccines are
conclusion. ineffective.' — Two people is
far too small.
COGENT Strong AND the premises are A strong argument where all
actually true. The inductive the observed facts are real and
equivalent of 'sound' in verified.
deductive logic.

Inductive vs. Deductive — Master Comparison

Feature Deductive Reasoning Inductive Reasoning
Direction General → Specific Specific → General
Certainty of conclusion 100% certain (if valid + true Probable — never 100%
premises) guaranteed
If all premises are true... Conclusion MUST be true Conclusion is LIKELY true —
but not certain
Can new evidence change No — the conclusion is logically Yes — new observations can
conclusion? fixed always overturn it
Quality terms used Valid / Invalid, Sound / Strong / Weak, Cogent /
Unsound Uncogent
Main use Maths, formal proofs, legal Science, research, everyday
arguments life, predictions
Famous example All humans are mortal → Every swan I've seen is white
 Socrates is mortal → All swans are white
Historical origin Aristotle (~350 BC) Francis Bacon (1600s)

Importance of Both Types of Logic

Both types of logic are essential — they complement each other rather than compete. Scientists use
inductive reasoning to form hypotheses from observations, then use deductive reasoning to test predictions
that must follow if the hypothesis is true. Lawyers use deductive reasoning to apply laws to cases, and
inductive reasoning to build arguments from evidence. Neither is superior — they serve different purposes.

💡 How They Work Together in Science

INDUCTIVE STEP: 'Every time we heat water to 100°C at sea level, it boils. Therefore, water
boils at 100°C at sea level.' → This forms a general hypothesis from observations.
DEDUCTIVE STEP: 'IF water boils at 100°C at sea level, THEN if I heat this water to 100°C at
sea level right now, it MUST boil.' → This predicts what must be true if the hypothesis is
correct. Both steps together = the scientific method.

Classifying Arguments — Practice

Here is how to tell whether an argument is inductive or deductive:

• DEDUCTIVE clue: The argument uses a general rule that applies to ALL members of a category, and
applies it to a specific case.
• INDUCTIVE clue: The argument uses past observations, patterns, or statistics to predict or generalise.

💡 Practice Examples from Lecture

1. 'The last six times we played our archrival in football, we won — so we'll win on Saturday.'
→ INDUCTIVE. It is based on past observations (6 previous wins) to predict a future outcome.
The win is probable but not certain. 2. 'There is no mail delivery on holidays. Tomorrow is
Thanksgiving. So the mail won't come.' → DEDUCTIVE. A general rule ('no delivery on
holidays') is applied to a specific case (Thanksgiving). The conclusion is certain. 3. 'All
vegetables are purple. Lettuce is a vegetable. Therefore lettuce is purple.' → DEDUCTIVE (but
UNSOUND, because P1 — 'all vegetables are purple' — is false). The structure is valid, but
the premise is wrong.

📝 QUICK RECAP
✅ Deductive logic: Aristotle | General → Specific | 100% certain conclusion
✅ Inductive logic: Francis Bacon | Specific → General | Probable conclusion — never certain
✅ Valid = correct logic structure. Sound = valid + true premises. (Deductive terms)
✅ Strong = well-supported by evidence. Cogent = strong + true premises. (Inductive terms)
✅ New evidence CAN overturn an inductive conclusion, but CANNOT overturn a valid
deductive one
✅ Both types are used together in science: induction to form hypotheses, deduction to test
them
❓ PRACTICE QUESTIONS
Q1. What is the key difference between deductive and inductive reasoning? Use one example
of each.
Q2. What does it mean for a deductive argument to be 'valid but unsound'? Give an original
example.
Q3. Is this deductive or inductive? 'Every student who studied with flashcards passed the
exam. I studied with flashcards. So I should pass.' Explain your answer.
Q4. What is the difference between a 'strong' and a 'cogent' inductive argument?
Q5. Who developed deductive logic and who developed inductive logic? What was each
person's key contribution?
 TOPIC 2: THREE KINDS OF INDUCTIVE ARGUMENTS
Now that we understand what inductive reasoning is, let's look at the three main types of inductive
arguments. Every inductive argument you encounter will be one of these three types: Generalisation,
Argument from Analogy, or Causal Inference. Each type works differently, has different strengths and
weaknesses, and is evaluated differently.

The Three Types at a Glance:

1. GENERALISATION → From a sample, draw a conclusion about the whole group
2. ANALOGY → Because two things are similar, they probably share another feature
3. CAUSAL INFERENCE → Event A regularly followed by B, so A probably causes B

Kind 1: Inductive Generalisation

📌 Inductive Generalisation
An inductive generalisation is when you observe a sample — a smaller group — and use what
you learn from that sample to draw a conclusion about the entire population — the larger
group. You are generalising from the part to the whole. The conclusion is probable, not certain,
because the sample might not perfectly represent the whole.

This is the most common type of inductive reasoning. It is used in surveys, polls, scientific studies, and
everyday observations. Every time you say 'most people think...', 'students generally prefer...', or 'research
shows that...', you are probably using an inductive generalisation based on a sample.

💡 Simple Example
You survey 200 university students and find that 80% prefer studying in the morning.
Conclusion: Most university students prefer studying in the morning. This conclusion is based
on a SAMPLE of 200 students and generalised to all university students (the POPULATION).
It is probable — but if the sample was only from one department, it might not represent all
students.

Key Terms in Generalisation

Target Population
📌 Target Population
The target population is the ENTIRE group that you want to draw a conclusion about — the
'whole' that your sample is supposed to represent. It is who or what your research findings will
be generalised to.

💡 Examples of Target Populations

→ 'All university students in Pakistan' → 'All diabetic patients in Karachi' → 'All voters in the
upcoming election' → 'All cars manufactured in 2020' Whatever general conclusion you draw
will apply to this entire group.
Sample
📌 Sample
A sample is a smaller subset of the target population that you actually study. Because studying
the entire population is usually impossible or too costly, you study a sample and use its results
to represent the whole. The sample must be representative — it must accurately reflect the
variety in the whole population.

💡 Example
Target population: All 10,000 students in a university. Sample: 200 students randomly
selected from all departments, years, and genders. The sample of 200 is used to represent all
10,000. If the sample is chosen well, the results should closely reflect the whole population.

Feature
📌 Feature
The feature is the specific characteristic, quality, or variable that you are measuring or
studying in your research. It is what you want to find out about the population.

💡 Examples of Features
→ Average monthly income of households → Preferred study time (morning / evening) →
Blood pressure levels in patients → Level of job satisfaction among employees →
Percentage of students who use social media for more than 3 hours daily

Related Factors
📌 Related Factors
Related factors are the background characteristics or conditions in the population that are
connected to the feature you are studying. They can influence the results. For a sample to be
representative, the related factors in the sample should be similar to those in the whole
population — otherwise the sample is biased.

💡 Example
Feature being studied: Job satisfaction among employees. Related factors: Age, experience
level, salary, type of work, gender. If you study job satisfaction but your sample only includes
young employees (age 20–25), you have not represented the related factor of age correctly —
and your results may not reflect the experience of older workers at all. Rule: Related factors in
your sample must mirror related factors in the whole population.

Biased Sample
📌 Biased Sample
A biased sample is a sample that does NOT accurately represent the target population. Some
groups are over-represented (included too much) or under-represented (included too little). A
biased sample produces misleading, unreliable, and unfair conclusions — because the sample
does not truly reflect the whole.

💡 Examples of Biased Samples

 → Surveying only private school students to find out what all students think about education
— public school students are completely excluded. → Asking only online users about national
issues — people without internet access are left out. → Studying employee satisfaction but
only asking managers — the views of junior staff are ignored. A biased sample always leads
to wrong generalisation conclusions — no matter how large the sample is.

Why Polls Can Be Unreliable

Polls are one of the most common uses of inductive generalisation. But polls can go wrong in several
ways. Understanding why polls fail is important for evaluating inductive generalisations critically.

Reason Polls Fail What It Means Example

Biased Sampling The sample doesn't represent Asking only online users about
the full population national issues leaves out
people without internet
Poor Question Wording Questions are leading, unclear, 'Do you agree the government
or emotionally loaded should stop wasting money?' —
already assumes waste
Respondent Dishonesty People may not answer Someone might say they
truthfully exercise daily when they don't
Respondent Confusion People don't fully understand Technical or complex questions
the question answered randomly to finish
quickly
Timing and Context Opinions change after recent A poll taken right after a
events disaster will show different
results than one a month later
Reporting Bias Media highlights only results A news channel only reports
that support their view the part of a poll that favours
their political position

Connection: Biased Sample and Related Factors

These two concepts are directly linked. A biased sample almost always happens when the related factors
in the sample are different from those in the target population. If a related factor like age, income, or
education is not properly represented in the sample, the sample becomes biased — and the generalisation
conclusion becomes unreliable.

💡 Connected Example
You want to study job satisfaction (Feature) among all employees in a company (Target
Population). Related factors: Age, salary level, department. If you only survey young
employees aged 20–25: → The related factor of AGE is not properly represented. → Your
sample is BIASED. → Your conclusion about job satisfaction across ALL employees is
UNRELIABLE. Fix: Make sure your sample includes employees of all ages, departments, and
salary levels.

Kind 2: Argument from Analogy

📌 Argument from Analogy
 An argument from analogy works by comparing two things. It says: 'Thing A and Thing B are
similar in several ways. Thing A has a certain feature X. Therefore, Thing B probably also has
feature X.' You are transferring what you know about one thing to another thing based on their
similarities. The conclusion is probable — it depends on how relevant the similarities are.

Analogies are used everywhere — in science, law, morality, history, and everyday conversation. When a
doctor says 'this treatment worked for similar patients,' or a lawyer says 'this case is similar to a previous
case,' or a scientist says 'this planet has conditions similar to Earth's,' they are all using arguments from
analogy.

💡 Basic Example
P1: A car and a motorcycle both have engines and wheels. P2: A car needs fuel (petrol) to
run. Conclusion: Therefore, a motorcycle probably also needs fuel to run. This is a
reasonable analogy — both vehicles are similar in relevant ways (both are engine-powered
road vehicles), so inferring a shared feature (needing fuel) is well-supported.

💡 Scientific Analogy Example

P1: Earth has an atmosphere, liquid water, and supports life. P2: Planet X has an
atmosphere and liquid water. Conclusion: Therefore, Planet X might also support life. This is
an inductive analogy — the conclusion is possible and probable given the similarities, but not
certain. Planet X could have other conditions that prevent life.

Key Terms in Analogical Arguments

Term Definition Example from 'Teacher is

like a Gardener'
Target Term The main subject you are trying Teacher (we want to
to explain or describe — what understand the teacher better)
the analogy is ABOUT
Comparison Term The thing you are comparing Gardener (used to shed light on
the target to — the analogy what a teacher does)
itself
Feature The specific quality or 'Helps others grow' — this is
characteristic being compared the feature that both the
— the shared property teacher and gardener share
Mentioned Similarities Similarities that are clearly and 'Both a teacher and a gardener
explicitly stated in the argument help living things grow' —
stated directly
Implied Similarities Similarities that are not stated Both nurture, both remove
directly but are expected to be obstacles (weeds / bad habits),
understood both need patience — implied
but not said

What Makes an Analogy Strong or Weak?

Not all analogies are equally good. The strength of an analogy depends on four key factors. The more
relevant the similarities, and the fewer the important differences, the stronger the analogical argument.
 Factor How It Affects Strength Example
Relevance of similarities Similarities must be logically A car and a motorbike sharing
connected to the conclusion — an engine is RELEVANT to
irrelevant similarities do nothing inferring both need fuel. Both
being manufactured in
Germany is IRRELEVANT to
their fuel needs.
Number of similarities More relevant similarities = If two watches share the same
stronger analogy manufacturer, materials, design
process, AND quality control —
the analogy that one is high
quality because the other is,
becomes very strong.
Number of differences More significant differences = If two hospitals are similar in
weaker analogy size but one is in a developed
country and the other in a
developing country —
differences in resources make
direct comparisons unreliable.
Diversity of cases If the analogy works across If a teaching method works in
many different contexts, it is primary schools, secondary
stronger schools, AND universities —
the analogy that it will also work
in your school is stronger.

💡 Weak Analogy Example

P1: A smartphone and a chocolate bar are both rectangular. Conclusion: Therefore, a
smartphone probably tastes sweet. This is a WEAK (and absurd) analogy. The shared feature
(rectangular shape) is completely IRRELEVANT to the conclusion (taste). Being the same
shape tells you nothing about flavour. Rule: Similarities must be RELEVANT to the feature
you are concluding about — otherwise the analogy is weak.

Uses of Analogy
Analogies are used for much more than just making arguments. Here are all the ways analogies appear in
reasoning:

Use What It Does Example

To Explain / Clarify Makes a difficult abstract idea 'The brain works like a
easier to understand by computer' — helps people
comparing it to something understand how the brain
familiar processes information
To Support Arguments Suggests that because two 'If Country A succeeded with
(Reasoning) things are similar in some this education policy, Country B
ways, they will be similar in — which has similar
another demographics — might also
succeed'
To Persuade Makes arguments more 'Just as a gardener must
emotionally convincing or remove weeds for flowers to
appealing grow, a teacher must correct
 bad habits for students to learn'
To Illustrate a Principle Shows how a general rule 'Managing a team is like
applies in a familiar situation steering a ship — you must
keep it balanced and on course'
To Discover New Ideas Finds patterns in one domain Doctors noticed the heart works
and applies them to another to like a mechanical pump → led
generate new theories to the invention of artificial
hearts and pacemakers

Analogy in Morality, Law, and History

Analogy in Morality
In moral reasoning, analogies help us decide what is right or wrong by comparing a new situation to one
where we already know the moral answer. If two situations are morally similar in all the relevant ways, the
same moral rule should apply to both. This is the principle of moral consistency.

💡 Moral Analogy Example

If it is wrong to kill one person to save five others when you push them off a bridge... Then it is
also wrong to remove one patient's organs to save five patients who need transplants. Both
situations involve deliberately killing one person to save several others. The morally relevant
features are the same, so the same moral judgment should apply to both.

Analogy in Law
Legal reasoning is built on analogy. Judges look at past court cases (called precedents) and ask: 'Is this
new case similar to that old case in the legally relevant ways?' If yes, the same legal ruling should apply.
This principle is called stare decisis — which is Latin for 'to stand by things decided.'

💡 Legal Analogy Example

Past case: A court ruled that posting someone's private photos online without consent is
invasion of privacy. New case: Someone shares private text messages of another person
online without consent. Legal analogy: Both cases involve sharing private, personal content
online without the person's knowledge or consent. The morally and legally relevant features
are the same. Conclusion: The same ruling — invasion of privacy — should likely apply to the
new case as well. This is how stare decisis works — past decisions guide future ones through
analogical reasoning.

Analogy in History
Historians use analogies to understand the present by comparing it to the past. If two historical situations
share important features, we can use what we know about how the old situation developed to make
predictions or draw lessons about the current one.

💡 Historical Analogy Example

Historians compare the Roman Empire and the British Empire to understand how empires rise
and fall. Both shared features: Vast territory, complex administration, military power,
economic dominance, and eventually overextension and decline. By studying how Rome fell,
historians can draw lessons about factors that may have contributed to the British Empire's
 decline — or about how modern powerful nations might face similar challenges.

Kind 3: Causal Inference

📌 Causal Inference
Causal inference is when you observe that Event A regularly happens before or alongside
Event B, and you conclude that A probably causes B. It is the process of establishing a cause-
and-effect relationship from observed patterns. The conclusion is inductive — probable but not
certain — because correlation (two things happening together) does not automatically prove
causation (one causing the other).

Causal inference is the foundation of medical research, social science, economics, and everyday problem-
solving. When researchers say 'smoking causes cancer' or 'exercise reduces depression' or 'poverty
increases crime rates,' these are all causal inferences built from large bodies of observational and
experimental evidence.

💡 Simple Example
Observation: Every time I study hard the night before an exam, I get a good grade.
Conclusion: Studying hard probably causes good grades. This is a causal inference. The
pattern is consistent and logical — studying hard gives you more knowledge, which leads to
better answers. The connection makes sense.

⚠️ The Key Danger: Correlation ≠ Causation

Just because two things happen together or one follows the other does NOT mean one
causes the other. This is the Post Hoc fallacy. Example: 'I wore my lucky shirt and we won the
match. My shirt caused the win.' → The win happened AFTER wearing the shirt — but that
doesn't mean the shirt CAUSED the win. Example: 'Countries with more television sets have
lower rates of malnutrition.' → This is a correlation — but TV sets don't prevent malnutrition.
Both are caused by a THIRD factor: wealth. Wealthier countries can afford both more TVs
AND better food. The correlation is real but the causal claim is wrong.

Comparison: All Three Kinds of Inductive Arguments

Feature Generalisation Analogy Causal Inference
Core idea From a sample, Because A and B are Event A regularly
conclude about a similar, B probably followed by B, so A
whole group shares A's feature probably causes B
Key question Is the sample large Are the similarities Is this a real cause, or
and representative? relevant and just a correlation?
numerous?
Main weakness Small or biased Irrelevant or Confusing correlation
sample insufficient similarities with causation
Example 200 students prefer Earth has water + life, Studying hard + good
morning study → all Planet X has water → grades → studying
students probably do Planet X may have life causes good grades
Common fallacy Hasty Generalisation Weak Analogy Post Hoc (False
Cause)
📝 QUICK RECAP
✅ Three kinds of inductive arguments: Generalisation, Analogy, Causal Inference
✅ Generalisation: sample → conclusion about whole population. Needs large, representative,
unbiased sample
✅ Key terms in Generalisation: Target population, Sample, Feature, Related factors, Biased
sample
✅ Analogy: A and B are similar → B probably shares A's feature. Strength depends on
relevance and number of similarities
✅ Key terms in Analogy: Target term, Comparison term, Feature, Mentioned/Implied
similarities
✅ Causal Inference: A regularly followed by B → A probably causes B. Danger: correlation ≠
causation
✅ Analogy used in morality (moral consistency), law (stare decisis / precedent), and history
(learning from past events)

❓ PRACTICE QUESTIONS
Q1. What is the difference between a target population and a sample? Why does the sample
need to be representative?
Q2. What is a biased sample? Give one original example and explain what related factor was
misrepresented.
Q3. Define 'argument from analogy.' What four factors determine whether an analogy is strong
or weak?
Q4. Identify the type of inductive argument: 'In every country where this economic policy was
introduced, GDP grew within 3 years. This country just introduced the same policy, so GDP
will probably grow.'
Q5. What is causal inference? Why is 'correlation does not equal causation' such an important
principle?
Q6. What is stare decisis? How does analogical reasoning make it work?
Q7. Evaluate this analogy: 'Countries are like children — they need firm discipline from a
stronger authority to develop properly.' What is strong or weak about this analogy?
 TOPIC 3: TOPICS OF INDUCTIVE REASONING
This topic covers three important statistical and methodological concepts used in inductive reasoning —
particularly in surveys, polls, and research. Understanding these concepts helps you evaluate how reliable
an inductive generalisation is. It also covers five common fallacies that appear specifically in inductive
reasoning.

1. Random Variation
📌 Random Variation
Random variation refers to the natural, unavoidable fluctuations that occur in data when you
repeat a process or measurement. Even when nothing has actually changed, the results of two
separate surveys or experiments on the same population will usually differ slightly — because
of unpredictable, uncontrollable factors. This is normal and expected. It does NOT mean
something went wrong.

Think of it this way: if you flip a fair coin 10 times, you expect about 5 heads. But you might get 6 or 7 —
not because the coin is unfair, but because of random variation. If you flip it 10 more times, you might get a
different result again. The more times you flip (the larger the sample), the closer you get to the true
probability of 50%.

💡 Example from Lecture

Survey 1: You ask 100 randomly selected students which subject they like best. Result: 40%
say Science. Survey 2: You ask a different 100 randomly selected students the same
question. Result: 37% say Science. The difference between 40% and 37% is NOT because
student preferences changed — it is random variation. Both surveys are estimating the true
preference of the whole student body, and small differences are expected. Key insight:
Random variation decreases as sample size increases. The bigger your sample, the more
reliable your result.

🔗 Connection to Generalisation
Random variation is why we can never be 100% certain about an inductive generalisation from
a sample. Even a perfect, unbiased sample will produce slightly different results than the
actual population — due to random variation. This is why we use margins of error and
confidence levels to acknowledge and quantify this uncertainty.

2. Margin of Error (Error Margin)

📌 Margin of Error
The margin of error tells you how much the survey result might differ from the true value in the
actual population. It is expressed as a ± (plus or minus) number, and it creates a range —
called a confidence interval — within which the true population value probably falls. A smaller
margin of error means a more precise result.

The margin of error exists because of random variation. Since we are studying a sample rather than the
whole population, our result will never be exactly right — there will always be some gap between the
sample result and the true population value. The margin of error tells us how large that gap might be.
 💡 Example
A poll finds that 60% of voters support Candidate A, with a margin of error of ±4%. This
means: The actual support in the real population is probably between 56% and 64%. → 60%
− 4% = 56% (the lower bound) → 60% + 4% = 64% (the upper bound) The true figure
might be anywhere in this range. If two candidates are separated by less than the margin of
error, the poll cannot reliably predict who will win.

📊 What Affects the Margin of Error?

→ LARGER sample size = SMALLER margin of error (more precise) → SMALLER sample
size = LARGER margin of error (less precise) Example: A poll of 100 people might have a
±10% margin of error. A poll of 10,000 people might have a ±1% margin of error. This is why
large samples are so important in research — they reduce the margin of error and make the
result more reliable.

3. Confidence Level
📌 Confidence Level
The confidence level tells you how sure you can be that the true population value actually falls
within the margin of error. It is expressed as a percentage — most commonly 90%, 95%, or
99%. A 95% confidence level is the most widely used standard in academic research and
polling.

The confidence level and the margin of error work together. Every poll or study uses both — you have to
know both numbers to fully understand how reliable a result is. A higher confidence level is more
reassuring, but it usually requires a larger sample size or a wider margin of error.

💡 Example
A survey has a 95% confidence level and a ±3% margin of error. What this means: If you ran
this same survey 100 times, approximately 95 of those 100 runs would produce a result within
±3% of the true population value. The remaining 5 out of 100 runs might produce a result
outside that range — due to random variation. So: We are 95% confident that the true value is
within our margin of error. We are NOT 100% certain.

📊 The Three Standard Confidence Levels

90% confidence level: Less certain. Used when a wider range of error is acceptable. 95%
confidence level: The standard in most academic and professional research. 99% confidence
level: Very high certainty. Requires a much larger sample size. Important trade-off: Higher
confidence level → requires larger sample OR wider margin of error. You cannot have 99%
confidence AND a tiny margin of error with a small sample — something has to give.

How All Three Concepts Connect

Concept What It Measures How They Are Related
Random Variation Natural fluctuations in data The reason why margins of
between samples error and confidence levels are
needed in the first place
 Margin of Error How far the sample result might Caused by random variation.
be from the true population Decreases as sample size
value (expressed as ±%) increases.
Confidence Level How certain we are that the A higher confidence level
true value falls within the requires a larger sample or a
margin of error (expressed as wider margin of error
%)

💡 All Three Together — One Example

A research team surveys 1,000 voters. Result: 55% say they will vote for Party A. Margin of
error: ±3% Confidence level: 95% Full interpretation: We are 95% confident that the true
proportion of voters who support Party A is between 52% and 58%. The ±3% exists because
of RANDOM VARIATION — even this well-conducted survey cannot match the true population
value exactly. If the true value were 52% and Party B had 48% support, Party A is likely
winning — but by a margin smaller than the margin of error makes us totally certain about.

Fallacies in Inductive Reasoning

A fallacy is an error in reasoning that makes an argument look stronger than it actually is. Inductive
fallacies specifically are errors that occur when conclusions are drawn from evidence that is too weak,
biased, or improperly interpreted. Here are the five main fallacies that appear specifically in inductive
reasoning.

1. Hasty Generalisation
📌 Hasty Generalisation
Drawing a broad conclusion about an entire group from a sample that is too small or not
representative enough. You are 'hasty' — rushing to a general conclusion before you have
gathered enough evidence.

💡 Example
'I met two people from City X and they were rude — so all people from City X must be rude.'
Two people out of a whole city is a tiny, non-representative sample. This is how stereotypes
and prejudice are born — hasty generalisations about entire groups based on a handful of
encounters.

🔗 Connection
This is the main fallacy of bad GENERALISATION arguments. A generalisation is only as
strong as its sample. Small sample = weak argument = risk of hasty generalisation.

2. False Cause (Post Hoc Fallacy)

📌 False Cause / Post Hoc
Assuming that because Event B happened after Event A, Event A must have caused Event B.
The name comes from Latin: post hoc ergo propter hoc — meaning 'after this, therefore
because of this.' It confuses sequence (one thing happening after another) with causation (one
thing causing another).
 💡 Example
'I wore my lucky socks, and our team won the game. My lucky socks must have caused the
win.' The win happened after wearing the socks — but there is no logical or physical
mechanism by which socks affect the outcome of a sports match. The relationship is pure
coincidence.

🔗 Connection
This is the main fallacy of bad CAUSAL INFERENCE arguments. Just because two events are
correlated in time does not mean one causes the other. You need more than a pattern — you
need a logical mechanism and controlled evidence to establish causation.

3. Slippery Slope
📌 Slippery Slope
Arguing that one small first step will inevitably lead to a chain of increasingly serious negative
consequences, without providing any evidence that this chain of events will actually happen. It
exaggerates the consequences of an action by claiming a domino effect.

💡 Example
'If we allow students to use calculators in class, next they won't learn basic arithmetic at all,
then they won't be able to think mathematically, and eventually they won't be able to think for
themselves at all.' Each step in this chain is possible — but the argument assumes all of them
will inevitably happen, without evidence. Many students use calculators and still develop
excellent mathematical and critical thinking skills.

🔗 Key Point
The slippery slope is not always a fallacy. If there is real, documented evidence that the chain
of events tends to happen (for example, research showing that early drug use is statistically
linked to later harder drug use), then it is a legitimate inductive causal claim. It is a fallacy
when the chain is assumed without evidence.

4. Weak Analogy
📌 Weak Analogy
Using an analogy where the two things being compared are not similar in the relevant ways
needed to support the conclusion. The shared features are irrelevant, insufficient, or too
superficial to justify the inference being made.

💡 Example
'Cars need fuel to run. Therefore, humans must also need fuel (petrol) to live.' Cars and
humans are NOT similar in the relevant ways. Cars have combustion engines; humans have
biological metabolisms. The comparison completely breaks down — the 'similarity' (both need
an energy source) is too superficial to support the specific conclusion (that the energy source
is the same).

🔗 Connection
This is the main fallacy of bad ANALOGY arguments. An analogical argument is only as strong
as the relevance of its similarities. If the similarities don't logically connect to what you are
concluding, the analogy is weak and the argument fails.
5. Anecdotal Evidence
📌 Anecdotal Evidence
Using a single personal story, isolated incident, or one person's experience instead of
systematic, large-scale data and research. Anecdotal evidence is appealing because it is vivid
and personal — but it is not a reliable basis for general conclusions because one person's
experience may be exceptional, unusual, or simply coincidental.

💡 Example
'My grandfather smoked his whole life and lived to 95 — so smoking can't be that bad for you.'
This ignores the overwhelming evidence from millions of cases studied over decades that
smoking dramatically increases the risk of lung cancer, heart disease, and other serious
conditions. The grandfather is a rare exception — and rare exceptions do not disprove general
patterns.

🔗 Connection
Anecdotal evidence is essentially a hasty generalisation from a sample size of ONE. It is
particularly deceptive because the personal story is emotionally compelling. Critical thinkers
learn to say: 'That's an interesting case — but what does the large-scale systematic evidence
show?'

Master Comparison: All 5 Inductive Fallacies

Fallacy Core Error Connected to Which Quick Example
Inductive Type?
Hasty Generalisation Too small or biased a Generalisation 'I met 2 rude people
sample used to reach from City X → all
a broad conclusion people from City X are
rude'
False Cause (Post Confusing sequence Causal Inference 'I wore lucky socks →
Hoc) with causation — after team won → socks
this, therefore because caused the win'
of this
Slippery Slope Assuming a small step Causal Inference 'Calculators in class →
inevitably leads to a students can't think →
chain of bad civilisation collapses'
consequences without
evidence
Weak Analogy Comparing things that Analogy 'Cars need petrol →
are not similar in the humans need petrol to
relevant ways live'
Anecdotal Evidence Using one personal Generalisation (sample 'My grandfather
story instead of of 1) smoked and lived to 95
systematic large-scale → smoking is safe'
evidence

📝 QUICK RECAP
✅ Random Variation: natural fluctuations in data — expected and unavoidable in any sample
✅ Margin of Error: the ± range within which the true population value probably falls
✅ Confidence Level: how sure we are (as a %) that the true value is within the margin of error
✅ Larger sample = smaller margin of error = more reliable result
✅ Hasty Generalisation: too-small sample → broad conclusion
✅ False Cause (Post Hoc): after this ≠ because of this
✅ Slippery Slope: assuming an inevitable chain of consequences without evidence
✅ Weak Analogy: comparing things not similar in the relevant ways
✅ Anecdotal Evidence: one personal story used instead of systematic large-scale evidence

❓ PRACTICE QUESTIONS
Q1. What is random variation? Why does it mean that no inductive generalisation can be
100% certain?
Q2. A poll shows 52% support for a policy with a ±4% margin of error and 95% confidence
level. What does this actually mean? Can you confidently say the majority supports the policy?
Q3. What is the difference between a 90% and a 99% confidence level? What trade-off is
involved?
Q4. Identify the fallacy: 'Every time the economy does well, the ruling party had a red party
colour. The red colour causes economic growth.'
Q5. Is this a slippery slope fallacy or a legitimate causal claim? 'Research shows that countries
that introduced strict financial regulation after banking crises consistently avoided future
banking crises. Therefore, introducing such regulation now is likely to prevent future crises.'
Explain.
Q6. What is anecdotal evidence? Why is it emotionally appealing but logically weak?
Q7. Give one original example of EACH of the five inductive fallacies and explain why each
one fails.
HOW EVERYTHING CONNECTS

The Big Picture of Inductive Reasoning:

Inductive reasoning is reasoning from evidence to probable conclusions. It is the main tool
of science, research, and everyday prediction.
It has THREE main forms: Generalisation (sample → whole), Analogy (similar things
share features), and Causal Inference (pattern → cause).
Every form can go wrong through fallacies: Hasty Generalisation, Post Hoc, Slippery
Slope, Weak Analogy, and Anecdotal Evidence.
We measure reliability of generalisations using: Random Variation, Margin of Error, and
Confidence Level.

If You See This... It Relates To... Critical Question to Ask

A survey result like '60% of Inductive Generalisation + How large was the sample?
people prefer X' Random Variation Was it representative? What is
the margin of error?
'±3% margin of error, 95% Margin of Error + Confidence What is the actual range? How
confidence level' Level confident can I truly be?
'A and B are similar, so B Argument from Analogy Are the similarities actually
probably has feature X' relevant? Are there important
differences?
'Every time A happens, B Causal Inference + Post Hoc Is this a real causal
follows — so A causes B' Fallacy mechanism, or just a
correlation?
'My friend tried it and it worked, Anecdotal Evidence Is this one person's experience,
so it works for everyone' or is there large-scale
systematic evidence?
'If we allow X, eventually Y and Slippery Slope (possible Is there evidence the chain of
Z will happen' fallacy) events actually tends to
happen?
'I met two people from Group X Hasty Generalisation Is two people enough to
who had trait Y — Group X conclude about an entire
must have trait Y' group?
Past legal case compared to Analogy in Law (Stare Decisis) Are the legally relevant features
current case truly similar, or just superficially
similar?

Good luck on your exam! You have got this 💪