Probability and statistical reasoning: QCE Philosophy and Reason
“Apply basic probabilistic reasoning to evaluate arguments, including conditional probability, base rates and common statistical fallacies”
Inductive strength is about probability: how far the premises raise the probability of the conclusion, running from 0 to 1. Conditional probability matters because P(A given B) can differ sharply from P(B given A), and the base rate (how common something is before evidence) must always be combined with new evidence, as the disease-test example shows. Good reasoning updates belief from the prior in proportion to how diagnostic the evidence is, the core of Bayesian thinking. Common statistical fallacies include the base-rate fallacy, the conjunction fallacy (the Linda problem), the gambler's fallacy, the prosecutor's fallacy and mistaking regression to the mean for a causal effect.
Jump to a section
What this dot point is asking
QCAA wants you to reason carefully about probability when evaluating arguments, because inductive strength is fundamentally about how evidence raises or lowers the probability of a conclusion. You need the idea of conditional probability, the crucial role of base rates, the basic logic of updating belief on evidence, and the common statistical fallacies that trap careful people. This sharpens the whole inductive strand.
The answer
Probability as degree of support
Inductive strength is a matter of probability: how likely the conclusion is, given the premises. Probabilities run from 0 (impossible) to 1 (certain). A key distinction is between the prior probability of a claim (before evidence) and its probability conditional on evidence. The probability of A given B, written P(A given B), can differ wildly from P(B given A), and confusing the two is a frequent reasoning error.
Base rates and why they matter
The base rate is how common something is in the population before you consider any specific evidence. Ignoring it produces the base-rate fallacy. Consider a disease that affects 1 in 1000 people and a test that is 99 percent accurate. If you test positive, your chance of actually having the disease is far below 99 percent, because the rare disease (low base rate) means most positives are false positives drawn from the huge healthy majority. The intuitive answer (99 percent) ignores the base rate; the correct answer is closer to 1 in 11. Doctors and jurors regularly get this wrong.
Updating on evidence
Good reasoning updates belief in proportion to evidence: start from the base rate, then revise as new evidence arrives, weighing how likely that evidence would be if the claim were true versus if it were false. This is the plain-language core of Bayesian reasoning, named after Thomas Bayes. The lesson for argument analysis: strong evidence is evidence that is much more likely if the conclusion is true than if it is false, and you must always combine it with the prior probability rather than judging the evidence in isolation.
Common statistical fallacies
- Base-rate fallacy: ignoring how common something is, as in the disease-test case.
- Conjunction fallacy: judging a conjunction more probable than one of its conjuncts. In the famous "Linda" problem studied by Kahneman and Tversky, people rated "Linda is a bank teller and a feminist" as more probable than "Linda is a bank teller," which is impossible, since a conjunction can never be more probable than either part.
- Gambler's fallacy: believing that independent events are owed a correction, for example that after five reds the roulette wheel is "due" black. Independent trials have no memory.
- Prosecutor's fallacy: confusing P(evidence given innocence) with P(innocence given evidence), making a small chance of a coincidental match sound like a small chance of innocence.
- Regression to the mean: extreme results tend to be followed by more average ones, which can be mistaken for a causal effect (for example, harsh criticism seeming to improve performance when the improvement was just regression).
Applying probabilistic reasoning
In a response, when an argument cites a statistic: (1) ask for the base rate, (2) check whether a conditional probability has been flipped, and (3) watch for the specific fallacies above. Numbers can make a weak argument look authoritative; probabilistic literacy is what lets you see through it.
A security scanner flags 95 per cent of genuine threats and falsely flags 2 per cent of harmless bags. About 1 in 5,000 bags contains a threat. A bag is flagged. How worried should we be?
Step 1: Choose a population and apply the base rate
Imagine 100,000 bags. With a 1 in 5,000 base rate, about 20 contain a threat and 99,980 are harmless.
Step 2: Count the true positives
Of the 20 genuine threats, the scanner flags 95 per cent, so about 19 true positives.
Step 3: Count the false positives
Of the 99,980 harmless bags, the scanner falsely flags 2 per cent, so about false positives.
Step 4: Compute the probability and interpret it
A flagged bag has probability of a real threat , under 1 per cent. Despite a "95 per cent accurate" scanner, almost every flag is a false alarm, because the threat base rate is tiny. This is why base rates must be combined with test accuracy.
A test that is 99 percent accurate does not make a positive result 99 percent likely to be true when the condition is rare. Because most people are healthy, the false positives from the large healthy group swamp the true positives. Ignoring the base rate is the single most common probabilistic error in real arguments.
- Flipping conditional probabilities
- P(positive given disease) is not the same as P(disease given positive); the prosecutor's fallacy makes the same flip.
- Ignoring base rates
- A high test accuracy can still yield mostly false positives when the condition is rare.
- The gambler's fallacy
- Independent events do not "remember" past results; a fair coin is never "due" heads.
- Rating a conjunction above its parts
- "A and B" can never be more probable than "A" alone, however plausible the extra detail sounds.
Try this
Q1. Explain the base-rate fallacy using a rare-disease test example. [4 marks]
- Cue. With a low base rate, most positives are false positives from the large healthy group, so a positive is far less than the test accuracy.
Q2. Why is "Linda is a bank teller and a feminist" never more probable than "Linda is a bank teller"? [2 marks]
- Cue. A conjunction cannot be more probable than one of its conjuncts.
Q3. Explain the gambler's fallacy. [2 marks]
- Cue. Independent trials have no memory; past results do not make a future independent outcome "due."
Exam-style questions
Questions in the style of QCAA exam questions on this dot point, each with a worked answer. They are written by ExamExplained unless tagged "Past paper"; the year shows the paper a question is modelled on.
Practice6 marksA disease affects 1 in 1,000 people. A test is 99 per cent accurate (it correctly identifies 99 per cent of those with the disease and gives a false positive for 1 per cent of those without it). A randomly tested person tests positive. Explain, with reasoning, why their probability of actually having the disease is far below 99 per cent.Show worked answer →
A 6 mark response works the base rates and contrasts true with false positives.
Set up 100,000 people. With a base rate of 1 in 1,000, about 100 have the disease and 99,900 do not.
True positives. Of the 100 with the disease, 99 per cent test positive, about 99 people.
False positives. Of the 99,900 without the disease, 1 per cent test positive, about 999 people.
Probability given a positive. Total positives are about . The chance a positive case truly has the disease is , about 9 per cent (roughly 1 in 11), not 99 per cent.
Why. The disease is rare (low base rate), so the false positives drawn from the huge healthy majority swamp the true positives. Ignoring the base rate is the base-rate fallacy.
Markers reward the true-versus-false-positive counts, the roughly 9 per cent result, and the base-rate explanation.
Practice4 marksExplain the conjunction fallacy using the "Linda" problem, and explain why the judgement it describes is necessarily mistaken.Show worked answer →
A 4 mark response states the fallacy and the probability law it breaks.
The fallacy. In the Linda problem (Kahneman and Tversky), people rate "Linda is a bank teller and a feminist" as more probable than "Linda is a bank teller". This is the conjunction fallacy: judging a conjunction more probable than one of its conjuncts.
Why it is necessarily mistaken. A conjunction (A and B) can never be more probable than either part on its own, because every case where A and B both hold is also a case where A holds, so . The vivid extra detail ("feminist") feels representative and inflates the perceived probability, but logically it can only reduce or hold it.
Markers reward the definition and the rule that a conjunction cannot exceed the probability of its conjuncts.