Non-clinical sciences/Stats & evidence · Bias & error

Confounding

[object Object]

What it means

A third variable associated with BOTH exposure and outcome and NOT on the causal pathway, distorting the apparent association. Address by design (randomisation, restriction, matching) or analysis (stratification, multivariable regression).

Worked example

The coffee–lung-cancer association is confounded by smoking (smokers drink more coffee and get more lung cancer).

In the exam

An apparent link between coffee and lung cancer disappears after adjusting for smoking, which is linked to both.

What settles it

Confounding is a third-variable distortion (fixable by adjustment); selection/recall bias arise from how subjects are chosen or remember (fixable mainly by design).

Classically confused with

Selection bias

Source: StatPearls