Question 1

What is a confounding variable?

Accepted Answer

A third factor associated with **both** the supposed cause and the supposed effect, creating an apparent relationship between them that is not causal — the central reason correlation does not establish causation.

**The classic illustration.** Ice cream sales correlate with drownings. Neither causes the other; **hot weather** causes both. Temperature is the confounder, and controlling for it makes the relationship disappear.

**Why this is so difficult in practice.** The confounder must be **identified and measured** to be controlled for, and the ones that matter are frequently the ones nobody thought of. Statistical adjustment can only account for variables you have data on — **residual confounding** from unmeasured factors remains, and it is the standard limitation of observational research.

**Real examples where confounding misled:**

**Coffee and lung cancer.** Early studies found an association. Coffee drinkers were more likely to smoke; **smoking** was the confounder.

**Hormone replacement therapy and heart disease.** Observational studies suggested a protective effect. Women receiving it were, on average, more affluent and healthier in other respects — **healthy user bias** — and randomised trials produced different results.

**Moderate drinking and health.** The apparent benefit is substantially complicated by the fact that abstainer groups include people who stopped drinking because they were unwell — the **sick quitter** problem.

**How confounding is addressed:**

**Randomisation**, which is the only method dealing with **unknown** confounders. Assigning treatment randomly means confounders, measured or not, are distributed equally between groups on average. This is why randomised trials sit high in evidence hierarchies — not because the analysis is better, but because the design removes a problem statistics cannot.

**Matching** participants on known factors.

**Statistical adjustment**, which works only for measured variables.

**Restriction** to a narrow group, at the cost of generalisability.

**Related concepts frequently confused:** a **mediator** sits on the causal pathway and should **not** be adjusted for if you want the total effect; a **collider** is a common effect of two variables, and adjusting for one can **create** a spurious association where none existed — a counterintuitive trap.

Question 2

What is statistical power, and why do underpowered studies mislead?

Accepted Answer

Power is the probability that a study will detect an effect if one genuinely exists. Low power means a real effect is likely to be missed — and, far less intuitively, it also makes the findings that

Question 3

What is a natural experiment?

Accepted Answer

A study exploiting a situation where circumstances have effectively assigned people to different conditions without the researcher doing it — allowing causal conclusions where a deliberate experiment

Question 4

What are some common chemistry mistakes to avoid?

Accepted Answer

Avoiding common chemistry mistakes is essential for accurate results and safety. Key areas include accurate measurements, safety protocols, and understanding reactions.

Question 5

What is regression to the mean?

Accepted Answer

The tendency for an extreme measurement to be followed by a less extreme one, simply because extreme values partly reflect chance — and failing to account for it produces confident conclusions about

Question 6

How does scientific consensus form, and what does it actually mean?

Accepted Answer

Through the accumulation of independent, mutually reinforcing evidence across many researchers and methods, until a position becomes the working assumption of a field — and it is a description of the

Question 7

What are the signs of climate change I should be aware of?

Accepted Answer

Signs of climate change include rising temperatures, melting ice, sea level rise, extreme weather events, changes in precipitation, ocean acidification, and ecosystem shifts. These indicators highlight the urgent need for action.

Related Questions