16.1 Inductive Logic, Generalisation & Causal Reasoning
Key Takeaways
- The NTS Analytical Reasoning curriculum names 'deductive and inductive logic' together, so inductive patterns are examinable alongside formal deduction.
- Deduction guarantees its conclusion when the premises hold; induction only makes its conclusion probable, which is why inductive arguments are attacked on sample quality.
- Correlation does not establish causation: the alternatives are reverse causation, a common third cause, and coincidence.
- Unrepresentative samples, self-selected samples and survivorship bias are the three sampling flaws NTS uses most often.
- A false analogy fails when the two cases differ on precisely the feature that drives the conclusion.
Inductive Logic, Generalisation & Causal Reasoning
The NTS curriculum describes the entire Analytical Reasoning segment in one line: "Deductive and inductive logic, Critical thinking and writing skills" — 30 questions on GAT-A and 20 on the other three categories. Formal deduction and puzzle diagramming are covered earlier in this guide. This section covers the other half of that line, inductive logic, which follows different rules and is attacked in different ways.
The Structural Difference
| Deductive | Inductive | |
|---|---|---|
| Direction | general rule → specific case | specific cases → general rule |
| Strength of conclusion | guaranteed if premises are true | probable only |
| Correct verdict labels | valid / invalid, sound / unsound | strong / weak, cogent / not cogent |
| How it is attacked | show the form is invalid | show the evidence is unrepresentative |
| Adding new evidence | cannot overturn a valid argument | can overturn a strong one |
Deductive: All GAT-A papers contain 35 verbal questions. This is a GAT-A paper. Therefore it contains 35 verbal questions. If the premises hold, the conclusion cannot fail.
Inductive: The last four candidates I met from this college scored above 70. Therefore candidates from this college generally score above 70. Even with true premises, the conclusion may be false — a fifth candidate could score 40.
This distinction decides which answer choices can be right. On an inductive stimulus, a choice claiming the conclusion "must be true" or "follows necessarily" is almost always wrong, because induction never delivers necessity.
The Four Inductive Patterns
1. Generalisation — from a sample to a population
In a survey of 300 candidates, 62% said they found the analytical segment hardest. Therefore most GAT candidates find it hardest.
Strength depends entirely on the sample. Three standard flaws:
| Flaw | Description | Example |
|---|---|---|
| Unrepresentative sample | the sample differs systematically from the population | surveying only candidates at one urban coaching centre |
| Self-selection | respondents chose to participate | an online poll answered only by the strongly opinionated |
| Survivorship bias | only successes remain visible | surveying admitted students about what preparation works |
Survivorship bias is the subtlest and the most heavily tested. Studying only those who succeeded tells you nothing about whether the same method failed for a hundred others who are no longer in the sample.
2. Statistical syllogism — from a population to an individual
Ninety per cent of candidates who complete three full mock papers finish the real test on time. Sara completed three mock papers, so she will probably finish on time.
The weakness is the unrepresentative individual: if Sara reads unusually slowly, the population statistic does not transfer.
3. Causal inference — from a correlation to a cause
The densest source of GAT items, treated in full below.
4. Analogical reasoning — from one case to a similar case
Metering tube wells reduced groundwater extraction in Province X, so it will reduce extraction in Province Y.
Strength depends on relevant similarity. A false analogy fails when the two cases differ on exactly the feature that drives the result — if Province Y's extraction is dominated by unmetered municipal supply rather than agriculture, the analogy collapses.
Causal Reasoning: The Four Explanations of Any Correlation
When a stimulus reports that A and B occur together and concludes that A causes B, exactly four explanations remain open:
- A causes B — the author's claim.
- B causes A — reverse causation.
- C causes both — a common third factor.
- Coincidence — no causal link, especially with a small sample.
Every weakening answer choice for a causal argument is one of options 2, 3 or 4. Every strengthening choice eliminates one of them.
Worked stimulus. Districts that opened new public libraries reported higher matriculation pass rates the following year. Therefore libraries improve examination results.
| Alternative | Concrete form |
|---|---|
| Reverse causation | districts with improving schools attract library funding |
| Common cause | rising district revenue funds both libraries and teacher salaries |
| Coincidence | pass rates rose province-wide that year for unrelated reasons |
Notice what a strengthener would look like: Neighbouring districts with comparable revenue but no new libraries showed no change in pass rates. That single sentence eliminates the common-cause and coincidence explanations at once, which is why controlled-comparison choices are so often correct.
Named Flaws Worth Recognising
| Flaw | What happens |
|---|---|
| Post hoc | treating "after" as "because of" |
| Hasty generalisation | a sweeping rule from too few cases |
| Sampling bias | the sample is not drawn from the target population |
| Confusing correlation with cause | ignoring alternatives 2, 3 and 4 above |
| False analogy | the cases differ where it matters |
| Appeal to authority | citing an authority outside their field |
| Ad hominem | attacking the arguer rather than the argument |
| False dilemma | offering two options where more exist |
| Circular reasoning | assuming what is to be proved |
| Equivocation | shifting the meaning of a key term mid-argument |
Equivocation deserves a note because NTS uses it in verbal-analytical crossover items. Only man is rational; no woman is a man; therefore no woman is rational fails because man means "human being" in the first premise and "male" in the second.
Working Method for Inductive Items
- Name the pattern. Sample-to-population, population-to-individual, correlation-to-cause, or case-to-case.
- Locate the leap. Every inductive argument jumps from evidence to a broader claim. Find the exact sentence where the scope widens.
- Attack the leap, not the evidence. The premises are given as true. A choice that disputes the data is out of scope.
- Match the flaw to the pattern. Generalisations fail on sampling; causal claims fail on alternative explanations; analogies fail on relevant difference.
- Reject necessity language. On inductive stimuli, "proves", "must", "guarantees" and "establishes conclusively" mark wrong answers.
The most common trap: a choice that is true, relevant, and strengthens an argument you were asked to weaken. Before comparing options, write a single letter beside the question — S or W — and check every surviving choice against it. Direction errors cost more marks on this topic than reasoning errors do.
A researcher finds that districts with more registered tube wells report lower average household incomes, and concludes that tube wells depress incomes. Which alternative explanation most directly undermines the causal claim?
A study surveys students who were admitted to MS programmes and finds that 85% used a particular preparation method, concluding the method is highly effective. What is the principal flaw?
Which feature distinguishes a strong inductive argument from a valid deductive one?
A columnist argues that because a congestion charge reduced traffic in a dense European city centre, the same charge will reduce traffic in a sprawling Pakistani city. Which response most effectively attacks the analogy?
You've completed this section
Continue exploring other exams