HSC Investigating Science Module 7 Fact or Fallacy: deep-dive 2026 guide

How Module 7 fits into HSC Investigating Science

Module 7, Fact or Fallacy, is where the methodology of Module 5 becomes a tool for judgement. Instead of designing investigations, you evaluate claims: is this scientific or pseudoscientific, is this correlation actually causation, is this evidence strong or weak, is this reasoning sound or fallacious. It is the module that most directly feeds the Section III extended response, where you are typically given a claim or scenario and asked to evaluate the science and the evidence.

The examiner reward pattern is consistent: criteria applied to specific named examples beat rote definitions every time. A student who writes "homeopathy is unfalsifiable because defenders attribute every failed trial to a non-individualised remedy, and dilutions of 10 to the power 60 leave no molecules of the original substance" outscores one who lists the eight features of pseudoscience without ever touching a real case. This guide builds the frameworks and pairs each with the named cases markers expect.

Distinguishing science from pseudoscience

Science is a process for producing knowledge through evidence and self-correction. Pseudoscience borrows the language and trappings of science without its substance.

Characteristics of a scientific claim:

• Falsifiable. A possible observation could prove it wrong.
• Empirical. Supported by observation or experiment.
• Peer reviewed. Scrutinised by independent experts before publication.
• Replicable. Independent researchers can repeat the method and obtain similar results.
• Provisional and self-correcting. Revised when new evidence emerges.
• Mechanistic. Offers a plausible cause linked to wider scientific knowledge.
• Quantitative. Makes specific, measurable predictions.

Characteristics of a pseudoscientific claim:

• Unfalsifiable. No conceivable observation could disprove it; contrary evidence is explained away.
• Anecdotal. Relies on testimonials rather than controlled studies.
• Not peer reviewed. Self-published and marketed directly.
• Not replicable. Methods are vague or proprietary.
• Resistant to revision. Defended even against overwhelming contrary evidence.
• No plausible mechanism. None proposed, or one inconsistent with established science.
• Vague. Qualitative claims like "balances energy" that are hard to test.
• Appeals to authority or tradition. Justified by celebrity or ancient wisdom rather than evidence.

The demarcation problem

The boundary between science and pseudoscience is debated by philosophers of science. Karl Popper (1934) proposed falsifiability as the criterion. Thomas Kuhn (1962) emphasised paradigms and normal-science problem-solving. Imre Lakatos (in the 1970s) distinguished progressive research programmes from degenerating ones. For HSC purposes, the Popperian falsifiability criterion plus the characteristics above is sufficient, but the strongest answers acknowledge that some claims sit at the boundary.

Boundary cases reward nuance. Acupuncture shows modest evidence for some pain conditions in controlled trials but weak or absent evidence for many other claimed indications, so it is partly evidence-based and partly marketed pseudoscientifically. Some supplement claims (correcting vitamin D deficiency) are evidence-based while others (anti-ageing) are not. The criteria let you reason about a claim piece by piece rather than labelling whole fields.

Correlation versus causation

A correlation is a statistical association; as one variable changes the other tends to change in a related way, measured by the correlation coefficient r from minus 1 to plus 1. Causation is a directed relationship where changing one variable produces a change in the other.

The asymmetry is the key fact: all causation produces correlation, but not all correlation indicates causation. Three reasons a correlation can exist without causation:

1. Confounding. A third variable causes both. Ice cream sales and drowning deaths both rise in summer because hot weather drives both. Eating ice cream does not cause drowning.
2. Reverse causation. The supposed effect is actually the cause. A correlation between cannabis use and depression might reflect people with depression self-medicating, rather than cannabis causing depression.
3. Chance. Random fluctuations produce associations in large datasets. Test 100 random hypotheses at p less than 0.05 and on average five will appear "significant" by chance alone.

The Bradford Hill criteria

In 1965 the epidemiologist Sir Austin Bradford Hill proposed considerations for inferring causation from observational evidence, used when randomised trials are unethical or impossible. They are a set of considerations to weigh, not a checklist:

• Strength of association. Strong associations are less easily explained by confounders.
• Consistency. Reproduced across populations, places and times.
• Specificity. The cause is linked to a specific outcome.
• Temporal sequence. The exposure precedes the outcome.
• Biological gradient (dose-response). More exposure produces more outcome.
• Plausibility. A mechanism consistent with current knowledge exists.
• Coherence. The relationship fits known facts about the disease.
• Experimental evidence. Where possible, intervention reduces the outcome.
• Analogy. Similar cause-effect relationships exist elsewhere.

The hierarchy of evidence

Not all evidence is equal. The hierarchy ranks study designs by how reliably they establish cause and effect, from strongest to weakest:

1. Systematic reviews and meta-analyses. Statistically combine all studies on a question. Cochrane reviews and NHMRC guidelines are the gold standard; they can detect small effects and quantify uncertainty.
2. Randomised controlled trials (RCTs). Random assignment, ideally double-blinded, balances confounders across groups so outcome differences can be attributed to the treatment.
3. Cohort studies. Groups followed over time; can detect associations but not establish causation alone. The 45 and Up Study (260,000 NSW participants) is an Australian example.
4. Case-control studies. Compare people with and without a condition, retrospectively; cheap but vulnerable to recall bias.
5. Cross-sectional studies. A snapshot at one time; cannot establish temporal sequence.
6. Case reports and case series. A single patient or small group; hypothesis-generating only.
7. Expert opinion and anecdote. Lowest; useful for context but not evidence of an effect.

Higher designs rule out more threats: meta-analyses and RCTs best control confounders, reverse causation and chance; case reports control none.

Evaluating a media claim

When a headline says "X causes Y", run a quick checklist: What study is cited, one study or a meta-analysis? What design, an RCT or a case report? What sample size? How large is the effect (effect size matters as much as statistical significance)? Has it been replicated? Who funded it? Most single observational studies overstate effects, which is why meta-analyses are needed to correct them.

Logical fallacies and cognitive bias

Faulty reasoning lets weak claims survive. Common fallacies to name in answers: post hoc ergo propter hoc (treating sequence as cause), appeal to authority (a claim is true because a famous person said so), appeal to nature (natural therefore safe), straw man (attacking a distorted version of an argument), and false dichotomy (presenting only two options). Cognitive biases include confirmation bias (seeking evidence that supports a prior belief) and the placebo effect plus regression to the mean, which together explain why people genuinely feel better after an ineffective remedy.

Check your knowledge

A mix of definitional, applied and evaluative questions covering this topic. Answer all under exam conditions, then check against the solutions block.

1. Explain the concept of falsifiability and why it is important for distinguishing science from pseudoscience. Include one example of an unfalsifiable claim. (4 marks)
2. Using one named scientific example and one named pseudoscientific example, distinguish between scientific and pseudoscientific claims. (6 marks)
3. Distinguish correlation from causation. Give one example of a correlation that is not causal and identify the confounder. (4 marks)
4. State four of the Bradford Hill criteria and explain how each supported the conclusion that smoking causes lung cancer. (5 marks)
5. Rank these designs from strongest to weakest evidence and justify the ranking: case report, cohort study, randomised controlled trial, systematic review. (4 marks)
6. A magazine reports that "eating chocolate prevents heart disease, according to a new study." Outline three questions you would ask before accepting the claim. (3 marks)
7. A claim notes that some children are diagnosed with autism within months of an MMR vaccination, concluding the vaccine causes autism. (a) Identify the logical fallacy. (b) Give the correct interpretation. (c) State one type of study that has refuted the claim. (4 marks)
8. A friend says a magnetic bracelet relieves their joint pain. (a) Give two reasons the claim fails scientific criteria. (b) Give one explanation for why the friend genuinely feels relief. (3 marks)